The Morphological Expression of Case In¨Ovdalian

Home , Elfdalian, North Germanic languages, Old Swedish

The Morphological Expression of Case in Ovdalian¨ ∗

Peter Svenonius CASTL, University of Tromsø May 16, 2014

Abstract The purpose of this paper is to document and describe the case system of the most conservative variety of Ovdalian¨ still spoken (‘Traditional Ovdalian,’¨ TO).¨ The system is compared with the four-case system of Old Swedish (OS) and the three-case system of Classical Ovdalian¨ (CO)¨ described by Levander (1909). I argue that TO¨ distinguishes three cases, but in full noun phrases, only manifests a two-case system, where Dative case is opposed to a Direct case comprising nominative and accusative functions. Pronouns generally show a different alignment, distinguishing nominative from objective cases. I focus on the nominal suffixes, which distinguish gender, number, definiteness, and case, as well as declension class. I argue that TO¨ nouns have only one suffix, where OS had three, and CO¨ was in transition between two suffixes and one. I examine the patterns of syncretism, and suggest that some can be explained in terms of markedness cooccurrence restrictions, but not others. I also briefly discuss adnominal modifiers (determiners and adjectives).

1 Introduction

Following the terminology adopted in Garbacz (2010) (which builds in turn on Helgander 1996 and other work), I will refer to the variety documented in detail by Levander (1909) as ‘Classical’ Ovdalian.¨ According to Levander (1925:37– 43), the most important distinguishing characteristics of Ovdalian¨ were present by the beginning of the 17th century, and possibly earlier. However, signiﬁcant changes took place in the 20th century. Following Garbacz (2010:34), a conservative variety of Ovdalian¨ spoken by a generation born after about 1920 but before 1950 can be called Traditional Ovdalian.¨ Most speakers today speak another variety, which is sometimes called ‘Younger’ Ovdalian¨ or ‘Modern’ Ovdalian.¨ 1

∗The endonym is (öv)dalska; Elfdalian, Ovdalian¨ , and Oevdalian have been variously used in works published in English. Thanks to Lars Steensland, NORMS, and Gerda Werf and Ulum Dalska and the other people of Alvdalen¨ for making the fieldwork possible and enjoyable. Thanks to Henrik Rosenkvist, Kristine Bentzen, and especially Piotr Garbacz for invaluable assistance during and after fieldwork. Lars Steensland, Piotr Garbacz, and an anonymous reviewer made many valuable suggestions and pointed out many errors in an earlier version, which I have endeavored to correct, but flaws no doubt remain, despite their efforts. 1According to a survey conducted in 2008 by Gösta Larsson, Ulla Welin, and Bengt Welin of the Ulum Dalska organization, approximately 2,500 people speak one variety or another of the language.

1 The Traditional Ovdalian¨ case system has not, to my knowledge, been systematically documented or described before. I describe it here and compare it to that of Classical Ovdalian.¨ I also locate both systems in the typological context of case systems cross-linguistically. Traditional Ovdalian¨ manifests a nominative-objective distinction in the pronominal system, but a direct-dative distinction on full nouns. Thus the language as a whole distinguishes three distinct syntactic case contexts, nominative, accusative, and dative, but the morphology of lexical noun phrases (including their determiners and modifiers) distinguishes at most two, direct versus dative. The Traditional Ovdalian¨ system reported here has been established mainly on the basis of a series of interviews and data elicitation sessions with speakers born before 1950 conducted by myself and associates in the NORMS project in 2007.2 Additional material was collected in follow-up expeditions by Piotr Garbacz and myself. In addition, the book Kunundsin kumb written by Hjalmar Larsson, a native speaker (listed in references under Larsson 1985; henceforth KK ). When quoting material from KK, I preserve the original spelling. For other material I conform to the standardized orthography established in 2005, except where deviations are phonologically significant and where otherwise noted.3 The symbol hDi represents a voiced interdental fricative /D/ in Classical Ovdalian,¨ wtih somewhat variable realization in different dialects of Traditional Ovdalian.¨ Additional sources are cited where used.4 I focus on full lexical nouns, discussing pronouns briefly in §2.2. The system of full lexical nouns shows two case forms in all three genders, in definite and plural forms, as well as some indefinite singular forms, see below. This is illustrated for the masculine noun est ‘horse’ in (1). (1) Masculine est ‘horse’ Singular Plural Indefinite Definite Direct est estn ester Dative est estem estum Descriptively, there are three significant differences between this system of noun inflection and the one described in Akerberg˚ (2000) and Dahl and Koptjevskaja- Tamm (2006). One is the absence of distinctions between the nominative and accusative cases in lexical nouns (compare Classical Ovdalian¨ ester (Nom) vs. esta (Acc) ‘horses’ in the indefinite plural, and estär (Nom) vs. esta˛ (Acc) in the definite plural, Levander 1909:11–12). The standard term for a single case which is used for both subjects and objects is Direct Case (Blake 2001, Haspelmath 2009), and I will use that term here. The second difference is the loss of definiteness distinctions in most plurals

2See http://norms.uit.no/ for information on the NORMS project. 3For example, Levander marks nasality with a hook on all nasal vowels, even where it is predictable (before nasal stops), whereas the standardized orthography marks nasality only when a nasal vowel is not followed by m or n, and that is the practice here, except for material quoted from KK, where nasality appears to be marked inconsistently. 4I have beneﬁtted from valuable reference materials for Ovdalian,¨ including a dictionary by Lars Steensland (2006) and a grammar by Bengt Akerberg˚ (2012). The latter only came out after this paper was completed, but I was able to consult earlier versions such as Akerberg˚ (2000). As beﬁts their function as reference works, these grammars are deliberately conservative, and I have not relied on them as evidence concerning how dative case is manifested in Traditional Ovdalian.¨

2 (again, compare the Classical Ovdalian¨ forms mentioned above: ester (Indef) vs. estär (Def) in the nominative, and esta (Indef) vs. esta˛ (Def) in the accusative). However, unlike the merger of the nominative and accusative, this is only true for some classes of noun. For example, a neuter noun like tak ‘roof’ shows a distinction between the definite and indefinite forms of the ‘direct case’ plural. (2) Neuter tak ‘roof’ Singular Plural Indefinite Definite Indefinite Definite Direct tak tatjeD tak tatje˛ Dative tak tatji˛ takum takum These two changes are likely to be related, since loss of distinct nominative and accusative forms eliminated part of the paradigm in which definite and indefinite forms were distinguished in the plural (the dative definite and indefinite plurals were already identical in Classical Ovdalian).¨ 5 The third difference in this system is the loss of several distinctly dative suffixes in the indefinite singular, with the result that Direct and Dative cases are often indistinct in the indefinite singular. This can be seen in the paradigms for est ‘horse’ and tak ‘roof’ in (1)–(2) above, as well as for the feminine noun fika ‘fig’ in (3) below.6 (3) Feminine fika ‘fig’ Singular Plural Indefinite Definite Direct fika fika˛ fikur Dative fika fikun fikum

Compare Classical Ovdalian¨ indefinite singular dative forms este ‘horse,’ tatji ‘roof,’ and fik˚a ‘fig’ (Levander 1909:11–12, 29–30, 40–41). Another interesting property in the system is a mismatch between the case system for lexical nouns and the case system for pronouns. First and second person pronouns, as well as plural pronouns, also make a two-case distinction, but syncretize accusative with dative rather than with nominative. The mismatch can be illustrated in the following way. (4) pronoun (1pl) lexical noun (str. m pl) Nom wi˛D kaller Nom Acc uoss kaller Acc Dat uoss kallum Dat ‘we/us’ ‘men’ A similar situation is observed in Norwegian dialects which preserve the dative. For example, Halsa (Nordmøre) dialect as described by Afarli˚ and Fjøsne (2012) distinguishes nominative from accusative/dative in parts of the pronominal sys-

5Levander (1909:13, n. 2) observes that older speakers occasionally have a distinct dative deﬁnite plural -ume, but this is suﬃciently marginal that he does not include it in his paradigms. 6I deliberately illustrate with a noun which is low in frequency in daily discourse. In elicitation sessions, I tested low as well as high frequency nouns to control for the possibility that case forms for certain high frequency nouns might be exceptional.

3 tem, but dative from nominative/accusative in some third person pronouns and on deﬁnite-marked lexical nouns, as indicated in (5). (5) pronoun (1pl) lexical noun (str. f sg) Nom me katt˚a Nom Acc ˚ass katt˚a Acc Dat ˚ass katt˚ainn Dat ‘we/us’ ‘the cat’ This kind of situation can be found in other case alignment systems. In ergative systems, the subject of an intransitive (S) shows the same case as the object of a transitive (O), distinct from the ergative case of the subject of a transitive verb (A), while in accusative systems S and A pattern together, distinct from O. Quite commonly, pronouns, and especially ﬁrst and second person pronouns, show accusative alignment while full nouns show ergative alignment (Silver- stein 1976, Dixon 1994). This is illustrated below with Dyirbal, an Australian language (Dixon 1972:42, 50). (6) pronoun (1pl) lexical noun (Class I sg) A (Nom) Nanad%i yaóa-Ngu A(Erg) S (Nom) Nanad%i yaóa S (Abs) O (Acc) Nanad%i-na yaóa O (Abs) ‘we/us’ ‘man’ Another similar pattern is the one seen in Bonan (Mongolic; Baerman 2009:226), where pronouns syncretize dative with accusative, while full nouns have a distinct dative, but have an accusative which is indistinct from the genitive. (7) pronoun ‘he’ lexical noun ‘foliage’ Nom ndˇzaN labˇcoN Nom Gen ndˇzaN-ne labˇcoN-ne Gen Acc ndˇzaN-de labˇcoN-ne Acc Dat ndˇzaN-de labˇcoN-de Dat

In both Dyirbal and Ovdalian,¨ there are a few points in the pronominal system where a three-way distinction can be seen. I discuss the facts for Ovdalian¨ in §2.2. More generally, the outline of the article is as follows. I compare Classical Ovdalian¨ to Old Norse in §2, stepping through the status of each of the cases in Classical Ovdalian.¨ In §3, I discuss the system of Traditional Ovdalian,¨ including the two-case system in full noun phrases. Because the system has not been carefully documented before, I take some time to establish that the system is in fact attested. There is a brief conclusion in §5.

2 From Old Norse to Classical Ovdalian¨

Family tree representations cannot easily depict the inﬂuence of languages on their neighbors, and cannot adequately reﬂect the complex status of Modern Norwegian, but to give a very rough idea of the place of Ovdalian¨ in the North Germanic language family, one might consider a representation such as the following.

4 2000 Icelandic Faroese Norwegian Danish Swedish Ovdalian¨ 1800 1600 Late Old 1400 Middle Middle Swedish Norwegian Danish 1200 Old Icelandic Early Old 1000 Old Danish Swedish 800 West Norse East Norse 600 400 200 North Germanic

Ovdalian¨ derives from Dalecarlian dialects which had already developed significantly distinctive properties from other Swedish varieties before the 1600’s (Levander 1925:37–43). The placement of Norwegian, Danish, and Swedish in the middle of the tree is meant to abstractly reflect the fact that those three languages have undergone a number of related developments which have not affected the graphically ‘peripheral’ languages (see e.g. Ringmar 2005 for a com- parison of conservative traits of Icelandic, Faroese, and Ovdalian).¨ In this section I place the Ovdalian¨ case system in its historical context, discussing the status of the genitive, the definite article, and some systematic syncretisms.

2.1 The four cases of Old Norse Old Norse7 had a four-case system with Nominative, Accusative, Dative, and Genitive. In the paradigms below, this is illustrated with examples from Old Swedish, from Noreen (1904) and Delsing (2002).8 Levander (1909) presents Classical Ovdalian¨ variably with four-case paradigms or with three-case paradigms, with the genitive left out. The many genitive gaps in his paradigms suggest that the genitive was already on its way out in the Classical Ovdalian¨ system. Sample paradigms are presented below as reported by Levander (but following the orthographic standard adopted in 2005 by the Ulum Dalska language council). (8) Paradigms for ‘horse’

7Here using the term Old Norse loosely to refer to the various medieval North Germanic varieties, including East Norse and West Norse. Because Old Norse is sometimes used more narrowly to refer to West Norse, Bandle et al. (2002) introduces the term Old Nordic, but this term is not yet widely known. 8The dual forms are not well attested in East Norse, but it is clear in West Norse that the dative and accusative are syncretized there (Noreen 1884, Iversen 1922).

5 Old Swedish Classical Ovdalian¨ M indef def M indef def sg N hestær hestrinn sg N est estn A hest hestinn A est estn D hesti hestinum D este estem G hests hestsins pl N hestar hestanir pl N ester est¨ar A hesta hestana A esta esta˛ D hestum hestumin D estum estum(e) G hesta hestanna (9) Paradigms for ﬁrst person pronouns Old Swedish Classical Ovdalian¨ 1 2 1 2 sg N iak þ¯u sg N ig du A mik þik A mig dig D m¯æ(r) þ¯æ(r) D mig dig G m¯ın þ¯ın du N vit it A oker iker D oker iker G okar ikar pl N v¯ı(r) ¯ı(r) pl N wi˛D ir A os iþer A uoss i˛D D os iþer D uoss i˛D G v¯ar iþar

Several considerations motivate treating the Ovdalian¨ genitive diﬀerently, as outlined in section 2.3 (modern colloquial Faroese has also essentially lost genitive as a case, Thr´ainsson et al. 2004).

2.2 Pronouns It can be seen from the pronominal forms in (9) that accusative is not distinct from dative in the dual and plural forms of the first person pronouns. The dative-accusative distinction is also lost for first and second person singular in the course of the Old Swedish period, with the historically accusative forms appearing in dative contexts (cf. Delsing 2002:929). The case which combines accusative and dative functions is sometimes called ‘objective,’ and this will be done here. Classical and Traditional Ovdalian,¨ like standard Swedish, exhibit a basic nominative-objective distinction through most of the first and second person. The distinction is illustrated for Swedish in (10a) and for Ovdalian¨ in (10b) (the example is elicited from a speaker of Traditional Ovdalian,¨ but can be used here in this discussion of characteristics of Classical Ovdalian¨ since the property in question is the same for both varieties). (10) a. Jag fr˚agar dig, ser du mig? I ask you.obj see you.nom me ‘I ask you, do you see me?’

6 b. Ig spyr dig, sir du mig? I ask you.obj see you.nom me ‘I ask you, do you see me?’ For such pronouns, dative is not distinguished from accusative, in Swedish (11a) nor in Ovdalian¨ (11b). (11) a. Jag g˚ar framför dig. I go before you.obj ‘I’ll go ahead of you’ b. Ig gor fromoni dig. I go before you.obj ‘I’ll go ahead of you’ This is the pattern for first singular, second singular, first plural, and third plural, as illustrated in (12). As in English, the second plural makes no case distinctions for nominative, accusative, and dative. (12) sg pl 12 123 nom ig du wi˛D iD dier obj mig dig uoss iD diem In the third plural, the old dative has taken over the role of the accusative, another trend which can be seen in Old Swedish (Delsing 2002:931). This change obliterates the accusative-dative distinction, just like the spread of the first and second person singular accusative to dative contexts. In Ovdalian,¨ one pronoun makes a three-way distinction, namely the feminine singular. The three forms are illustrated below with examples from KK, with two examples of the dative to show two alternate forms. (13) a. o˛ add it ba riskirad p˚ajkan, o˛ add riskirad estn og she had not only risked the.boy she had risked the.horse too ‘She had not only risked the boy, she had risked the horse as well’ b. Se add an o˛n ta leso etter uord fer uord then had he her.acc to read after word for word ‘Then he had her read after him, word by word’ c. Fast ieder kelindse wa so gomol so e’ wart fel it although one.of the.woman was so old so it was you.know not just so mitsid gart fer enner. just so much done for her.dat ‘On the other hand, one of the women was so old that there really wasn’t all that much that could be done for her’ d. so an sagd ad ien i fläd sett, at dsjäwo enn ien so he said to one in retinue his to give her.dat a caroliner. Caroliner ‘so he said to one of the men in his retinue to give her a Caroliner (a coin)’ The masculine singular pronoun shows the same split between direct and dative case that is seen in the nominal paradigm, as illustrated below (two variants of

7 the dative form are shown; the direct-case form an is variously glossed ‘he’ or ‘him’ here as elsewhere). (14) a. An war iend p˚ajkan i ien sturan krippuop i he was only the.boy in a big child.bunch in Spajk-gardem. The.Nail-Farm.dat ‘He was the only boy in a big bunch of kids on the Spajk (‘Nail’) farm’ b. Dier add aft Luok-Marit ta ˚aw˚a an i syn iel tide˛ they had had Luok-Marit to have him in sight all the.time ‘They had had Luok-Marit keep him in sight at all times’ c. Ed add itt gaj fo˛ no’ liwstecken fro˛ o˛m it had not gone get any life.sign from him.dat ‘It hadn’t been possible to get any sign of life out of him’ d. Fer o˛num add e’ wiktugestad we’ ta redd estn, e’ for him.dat had it most.important been to save the.horse the dyrestad so fanns i iel gardem. most.valuable as was.found in whole the.farm.dat ‘For him, the most important thing had been to save the horse, the most valuable thing there was on the whole farm’ Third person singular pronouns can be presented as in (15), arranging the genders in decreasing order of presumed markedness. (15) F M N nom o˛ an eD acc ona,on an eD dat enner onum, om dy¨o Thus, although the pronominal system seen as a whole makes a three-way case distinction, it is only in the third person singular forms that anything like a specialized dative can be identiﬁed. As mentioned in §1, it is not uncommon for languages to at least partly dissociate the case system expressed by pronouns from the case system expressed by full noun phrases. For instance, English has a two-case system for pronouns but no vestige whatsoever of that distinction in the case system. It seems that for Ovdalian¨ we can identify two competing case systems, as represented in (16) (cf. (4) in §1). (16) 1, 2, and pl pronouns 3sg m/n pns, lexical nouns Nom ig dier an kaller buordeD Drct Obj mig diem onum kallum buorde˛ Dat ‘I/me’ ‘they/them’ ‘he/him’ ‘men’ ‘the table’

Compare the corresponding elements in Early Old Swedish in (17). (17) N iak þ¯er han karlar borDit A mik þ¯a han karla D m¯ær þ¯em honum karlum borDinu ‘I/me’ ‘they/them (m)’ ‘he/him’ ‘men’ ‘the table’

8 Because of the general tendency for a nominative-accusative distinction to be made, the failure of the third person singular masculine pronoun to do so might be an accidental syncretism, rather than anything systematic (compare Gothic third singular masculine nominative is, accusative ina, dative imma, genitive is, with an apparently accidental nominative-genitive syncretism). However, the nominative-accusative syncretism in borD ‘table’ is systematic, for neuter, in all declension classes, both numbers, and so on. As already noted, the pronominal system was already losing dative-accusative distinctions in Late Old Swedish, as indicated in (18), which might be seenas a transitional point between Early Old Swedish and the Classical Ovdalian¨ system.9 (18) N iak þ¯er han karlar borDit A han karla mik þ¯em D honum karlum borDinu ‘I/me’ ‘they/them (m)’ ‘he/him’ ‘men’ ‘the table’ The tendency here might be related to Differential object marking, or DOM. DOM systems often make nominative-accusative distinctions only at the high end of a kind of ‘individuation’ hierarchy, for example animates or definites may be overtly marked for accusative case, while inanimates or indefinites fail to show overt marking (Bossong 1991, Dalrymple and Nikolaeva 2011). First and second person are the most highly ‘individuated,’ being inherently animate and definite. Silverstein (1976) posits an ‘animacy’ hierarchy of nominal types for case- marking cross-linguistically (drawing evidence from several Australian languages as well as Chinookan). A version of the hierarchy is presented in (19) (cf. Sil- verstein 1976:121-122; Dixon 1994:88-90 argues for 1st > 2nd, but for reasons stated below I put them into a single class). (19) 1st, 2nd > 3rd pn > Proper N > human N > animate N > inanimate N Silverstein’s observation is that if a language overtly marks accusative case on any class of noun phrases in this hierarchy, it will do so for those noun phrases which are higher than that class on the hierarchy. Thus, for example, Dyirbal only marks accusative case on first and second person pronouns, while another Australian language, Bandjalang, marks accusative on all pronouns, and another, Aranda, marks pronouns and animate nouns. If we equate neuter gender with inanimacy (at least abstractly, or historically), then we can say that Old Norse obeys this hierarchy, since it marks accusative on all types down to and including animate N (i.e. masculine and feminine, but not neuter). I argue below that Traditional Ovdalian¨ has lost accusative marking on all common

9I continue to indicate the nominative-accusative han-han syncretism as accidental. This is on the assumption (quite possibly wrong) that a systematic syncretism domain always respects the Silverstein hierarchy. The third person singular is separated from the more general domain of that syncretism in the neuter gender by the lexical masculine nouns. The diﬀerence with (16) is that if feminine singular pronouns and plural pronouns are treated as higher than masculine singular ones on the Silverstein hierarchy, then it is possible to posit a contiguous Direct-Dative syncretism domain including the masculine singular pronouns. Note that in modern standard Swedish, honum (now honom) has taken over accusative functions, thus patterning with the upper end of the Silverstein hierarchy.

9 nouns, preserving it only on first and second person pronouns, and some third person pronouns (feminine and plural). This is again perfectly compatible with the Silverstein hierarchy. Silverstein and Dixon also discuss cases where number matters, as in Ovdalian,¨ though they interpret the data differently. A general tendency that seems to hold for the cases here is that if a class in the hierarchy is split, then marked members of the class (e.g. plural, or feminine) tend to pattern toward the top of the scale while unmarked features tend to pattern toward the bottom (stated generally by Silverstein in terms of ‘positive’ values of features; Dixon instead suggests (p. 92) that number is not part of the hierarchy, and that plural forms will independently tend to make fewer distinctions than singular forms). Silverstein and Dixon relate the animacy hierarchy to the likelihood of a class of noun phrase acting as an agent or patient; noun phrases higher on the hierarchy are more typical agents, those lower are more typical patients. They show that the same hierarchy governs the distribution of ergative case-marking: if a language marks ergative on a type of noun phrase, it will do so on noun phrases lower than it on the hierarchy (see also e.g. DeLancey 1981). Here, we see that the same hierarchy appears to relate to dative. In Old Norse, the dative-accusative distinction is made consistently at the low end of the hierarchy, and fails to be made only at the very top, in part of the first and second person paradigm (namely the plurals; notice that this pattern fits the hierarchy as stated in (19), with 1st and 2nd combined into a single ‘participant’ category, but not Silverstein’s or Dixon’s versions). In Ovdalian,¨ the dative-accusative distinction is still made at the low end of the hierarchy, but the cut-off is lower than it was in Old Norse, somewhere in the third person pronominal system (again with plural above the cut, but here also the feminine singular). Pronominal systems and determiner systems are often related, and in Classi- cal Ovdalian¨ this relationship can clearly be seen. An interesting question which arises is whether the system of determiners in Traditional Ovdalian¨ follows the Nom-Obj alignment seen on the left in (16), or the Drct-Dat alignment seen on the right. In section 3.6, I show that the determiner system, to the extent that it expresses case, follows the Drct-Dat alignment. This supports the suggestion that what has occurred is a systematic structural change in the case system, rather than simply the accumulation of phonological changes; if the loss of case distinctions were entirely due to phonological leveling, then there would be no reason for the adnominal modifiers to level to a pattern different from what is seen in the pronominal system.

2.3 Genitive Most of the case forms are transparently related to their cousins in other Ger- manic languages, such as Icelandic, but the genitive is anomalous. Levander (1909) often gives only three cases in his nominal paradigms, or gives genitive forms only for deﬁnite-marked nouns. Dahl and Koptjevskaja-Tamm (2006) argue that Classical Ovdalian¨ is really a three-case system, and I adopt their conclusion here. I summarize their argument in this subsection and add a few observations. First, the existence of possessive forms of pronouns such as my in English is not suﬃcient evidence for a genitive case; such elements may be separated

10 entirely from the case system. In Old Norse, a genitive form of the pronoun can be used possessively but also shows up in all other genitive contexts, for example as complement of the preposition til ‘to’, which idiosyncratically assigns genitive: til m´ın, ‘to me.gen’. Possessive pronouns like English my do not show this wider distribution, and nor do the Ovdalian¨ possessive pronouns, apparently. Another reason to question the existence of genitive as a living case in Ovdalian¨ comes from the inflectional paradigms of nominal modifiers. In Old Norse, nominal modifiers normally agree in case with the noun phrase they modify, and accordingly display a full four-case paradigm; this holds of demonstratives, quantifiers, numerals, adjectives, and so on. For Classical Ovdalian,¨ however, Levander usually gives three-case paradigms, only exceptionally noting a genitive form (e.g. miklumes ‘many’ on p. 54). In Traditional Ovdalian,¨ such exceptional genitive forms of nominal modifiers are if anything rarer and more restricted. A third argument that there is no genitive case in Ovdalian¨ comes from the syntactic distribution of forms. In Old Norse, the genitive appears in a variety of contexts, including various adnominal uses as well as on complements of certain verbs and adjectives and prepositions. In Classical Ovdalian,¨ we can divide the potential genitive contexts into two, which turn out to behave differently: one is in the complement of certain prepositions, discussed in section 2.3.1, and the other is an adnominal possessive, discussed in section 2.3.2 below.

2.3.1 Distribution of genitive DPs; Prepositions In Ovdalian,¨ there is a large number of expressions in which a genitive form of a noun is combined with et ‘to’ or i ‘in’, for example (Levander 1909:96) et bys ‘to town,’ et endes “to end” meaning ‘in the end, finally,’ i kwelds “in evening” meaning ‘yesterday evening,’ etc. The -s ending is historically the regular genitive marker for singular masculine and neuter nouns of the strong declension, but some examples show historically plural, feminine, or weak endings other than -s, e.g. et juoler ‘for Christmas.’ Swedish, too, retains a number of such expressions, for example till skogs/till havs/till fjälls, ‘to the woods/to the sea/to the mountains’ including some which preserve old weak and feminine forms, e.g. till handa ‘at hand’, till salu ‘for sale’.10 In Ovdalian¨ as in Swedish, the collocations with prepositions do not allow free modification, determination, or quantification of the noun. English similarly has many collocations of P+N which do not allow modification; consider at hand ‘available’, on time ‘punctual’, by boat ‘using a boat as means of transport’, which do not allow modification or quantification (They arrived by boat, They arrived by speedboat, *They arrived by speedy boat, *They arrived by two boats). Unlike Swedish and Ovdalian,¨ English does not seem to have preserved any case endings in such collocations. Occasionally, such a construction allows some limited productivity, consider English by car, by plane, by hydrofoil. There must have been a productive stage for some part of this construction at some point in the history of Ovdalian¨ as well, since Old Norse ´ı ‘in’ does not assign genitive.

10Thanks to Lars-Olof Delsing for discussion, and for directing me to the informative entry for till in Svenska Akademiens Ordbok.

11 At this point, however, the productivity of the -s ending in these collocations in Ovdalian¨ is presumably limited at best. Thus, these constructions, though interesting in their own right, do not provide evidence for a full-ﬂedged genitive case in Classical Ovdalian¨ on a par with the dative.

2.3.2 The possessive Genitive is a general adnominal case in Old Norse (e.g. nafn V´ıga-Hrapps “name V´ıga-Hrapp.gen”, ‘the name V´ıga-Hrapp’; eyrir brends silfrs “ounce pure.gen silver.gen”, ‘an ounce of pure silver’; vika sjóvar “week sea.gen” ‘a sea-mile’; Nygaard 1966). The possessive use can be considered an example of this. (20) skáld Haralds hins hárfagra skald Harald.gen the.gen hair.fair.gen ‘Harald the Fair-haired’s skald (bard, poet)’ (Old Norse) As can be seen from the example, genitive case appears on the head of the possessor as well as the modifiers, in Old Norse.11 In Ovdalian,¨ there is a possessive construction which is clearly related historically to the genitive, in employing a morpheme with an /s/ in it, but diverging significantly from the Old Norse construction. (21) IttaD-jär ir kullumes saing. this-here is girl.dat.pl.poss bed ‘This is the girls’ bed’ (Classical Ovdalian,¨ Levander 1909:96) There are several differences here. One is that the /s/ genitive is historically restricted to strong masculine and neuter nouns, and is not found with feminine nouns. Another is that the /um/ is identical with the dative, as if the possessive form were built on top of the dative (as suggested in the gloss). Another is the prenominal position of the possessor, which was not the unmarked order in Old Norse (cf. (20) and e.g. Nygaard 1966:129, Faarlund 2004:59). If the -es is a possessor marker, rather than a case, then that would partly explain why Levander (1909) often only gave ‘genitive’ forms for definite-marked nouns: possessors are typically definite, and the possessor construction might even require definiteness. An important difference comes through when postnominal modifiers are added: the possessive marking here does not show up on the head, but rather the periphery of the phrase (as with the English s-possessive), here set off with a hyphen.12 (22) Ann upp i buDum-es etta Ann up in summer.pastures.dat.pl-poss hood ‘Ann at the summer pastures’ hood’ (Classical Ovdalian,¨ Levander 1909:97)

11Though the form fagra ‘fair’ is in the weak declension, because of the deﬁnite article, and therefore fails to distinguish genitive from dative and accusative. 12BuD ‘summer pasture’ is in the dative plural here, dative because of the preposition; Ann shows no overt case ending and can be assumed to be nominative; a dative or accusative form would be Anno, cf. Levander (1909:36).

12 Dahl and Koptjevskaja-Tamm (2006), observing these and other examples, con- clude that the /es/ is a possessive clitic attaching outside the dative, as suggested by the gloss. The relevant parts of the paradigm for masculine ‘horse’ and feminine ‘girl’ are given in (23) (again, using a hyphen to graphically distinguish the possessive marker, and now introducing boxes to mark systematic syncretisms). (23) indef def indef def sg D este estem sg D kullu kullun G est-es estem-es G kull-es kullun-es pl D estum pl D kullum G estum-es G kullum-es The clitic analysis explains several things: the morphological complexity (i.e. the fact that the dative morphology cooccurs with the genitive, unlike the situation in Old Norse), the insensitivity of the -es form to gender and declension classes (again, unlike the Old Norse genitive), the paradigm gaps (the possessive construction which licenses the clitic may require a deﬁnite possessor), and more (see Dahl and Koptjevskaja-Tamm 2006 for additional arguments).

2.4 Nominative-Accusative The Old Norse paradigms showed some syncretisms for nominative and accusative case, as illustrated here with one example each of a strong declension masculine, strong declension feminine, and strong declension neuter noun (in Old Swedish). Again, the boxes indicate systematic syncretisms for adjacent case-cells.13 (24) The case/number paradigms for a strong masculine noun: ‘horse’ (M) indeﬁnite deﬁnite sg nom hestær hestr-inn acc hest hest-inn dat hesti hesti-num gen hests hests-ins pl nom hestar hesta-nir acc hesta hesta-na dat hestum hestum-in gen hesta hesta-nna (25) The case/number paradigms for a strong feminine noun: ‘journey’ (F)

13For the strong feminine declension, the nominative, accusative, and dative are usually the same, as here, but not always: e.g. graf ‘grave.nom/acc’, gravu ‘grave.dat’; k´yr ‘cow.nom’ k´u ‘cow.acc/dat’. Similarly, the apparent accusative-genitive syncretism in the strong masculine plural paradigm is not systematic: e.g. gesti ‘guests.acc’ vs. gesta ‘guests.gen’. Thus the boxes in the examples only mark exceptionless syncretisms.

13 indefinite definite sg nom færþ færþ-in acc færþ færþ-ina dat færþ færþ-inni gen færþar færþ-innar pl nom færþir færþi-nar acc dat færþum færþum-in gen færþa færþ-inna (26) The case/number paradigms for a strong neuter noun: ‘ship’ (N) indefinite definite sg nom skip skip-it acc dat skipi skipi-nu gen skips skips-ins pl nom skip skip-in acc dat skipum skipum-in gen skipa skipa-nna As indicated, nominative and accusative are fully syncretized for the neuter and the feminine plural, but not systematically for the masculine nor for the feminine singular. For weak nouns, the same nominative-accusative syncretisms are observed for the neuter and the feminine plural, and in addition there are systematic syncretisms for the non-nominative cases (Acc, Dat, Gen) in the indefinite singular. (27) The case/number paradigms for a weak masculine noun: ‘bull’ (M) indefinite definite sg nom uxi uxi-nn acc uxa-nn dat uxa uxa-num gen uxa-ns pl nom uxar uxa-nir acc uxa uxa-na dat uxum uxum-in gen uxa uxa-nna (28) The case/number paradigms for a weak feminine noun: ‘fly’ (F) indefinite definite sg nom fluga fluga-n acc flugu-na dat flugu flugu-nni gen flugu-nnar pl nom flugur flugu-nar acc dat flugum flugum-in gen flugna flugna-nna (29) The case/number paradigms for a weak neuter noun: ‘ear’ (N)

14 indefinite definite sg nom øra-t acc øra dat øra-nu gen øra-ns pl nom øron øron-in acc dat ørom ørom-in gen øra øra-nna There are further syncretisms which are more restricted in scope. For example Old Norse strong masculine nouns ending in liquids such as /l/ (such as karl, ‘man’) neutralized the nominative-accusative distinction in the singular.14 This is not boxed here as it does not affect the entire strong masculine declension, at least not in the early part of the Old Swedish period. (30) Paradigm for ‘man’ Old Swedish M indef def sg N karl karlinn A karl karlinn D karli karlinum G karls karlsins pl N karlar karlarnir A karla karlana D karlum karlumin G karla karlanna Another systematic syncretism which can be observed in the paradigms above is that the genitive plural -a and the dative plural -um do not distinguish gender. Genitive, Dative, and plural are all traditionally marked categories, and such syncretisms can be described in terms of Impoverishment (Bonet 1991, Noyer 1992, Halle 1997), a kind of redundancy rule which states the incompatibility of one marked feature with another. Such rules can be stated as cooccurrence restrictions on marked features (e.g. *[gender,plural] if gender distinctions are absent in the plural), but are normally stated in Distributed Morphology in terms of the deletion of a marked feature.15 Neuter pronouns are arguably the least-marked; they are used as expletives, and to refer to clausal referents which have no gender (and in Old Norse, a coordination of a masculine and a feminine controls neuter plural agreement, cf. Faarlund 2004:95). If we assume that neuter is the absence of gender features, then an Impoverishment rule can delete masculine and feminine in the context of genitive plural or dative plural, as in (31), letting anim[ate] stand in as a generalization over masculine and feminine (see Müller 2005 for a more detailed account of Impoverishment rules in Icelandic nouns).

14Through assimilation of /r/ followed by word-ﬁnal degemination, Noreen (1904:182, 282). 15As noted by Carstairs (1987), neutralizations are more common in portmanteaux morphemes than in fully agglutinative systems. Thus the cooccurrence restriction could be seen as a restriction on the speciﬁcation of an exponent, though this is not how it is modeled in Distributed Morphology.

15 (31) a. anim → ∅ / Gen,pl b. anim → ∅ / Dat,pl Stump (2001:238), developing a suggestion by Hjelmslev (1935), proposes a Fea- ture Ranking Principle which states that syncretisms are governed by language- specific feature hierarchies. (32) Feature Ranking Principle: For any language ℓ, there is a ranking > of morphosyntactic features in ℓ which satisfies the following condition: for every stipulated syncretism S in ℓ, if the dominant properties of S include a specification of the features Fd and the subordinate properties of S include a specification of the feature Fs, then Fd > Fs. If case and number are ranked higher than gender in Old Swedish, then the Feature Ranking Principle would allow the syncretisms described by (31a) and (31b). The ranking of case and number above gender fits with their locus in an extended DP (Borer 2005, Svenonius 2008): case values are determined ex- ternally to the noun phrase, so it is reasonable that case is structurally higher than DP-internal features. Gender is lexically determined, so it is reasonable that it should be structurally lower than number. It is plausible that the structural hierarchy K[ase] > Number > Gender might be reflected in the Feature Ranking Principle. Suppose that expressing features on an exponent is ‘costly’; expressing two features is better than expressing three. Of the three features case, number, and gender, gender is lowest on the hierarchy and hence the most likely to be deleted as in (31a) and (31b). There is also a systematic syncretism of nominative and accusative in the feminine paradigms. If cases are organized in a markedness hierarchy, so that Nominative is the absence of Case, and Accusative is the presence of Case but the absence of any Oblique case (essentially as in Jakobson 1936; see Caha 2009 for a recent account), then the systematic syncretism of nominative and accusative in the feminine plural could be stated as follows (using Acc for the relevant case component). (33) Acc → ∅ / fem,pl If gender and number are ranked higher than case, then (33) would be compatible with the Feature Ranking Principle. But to allow it as well as (31a) and (31b), the ranking would have to place accusative over gender (at least feminine), and gender (including masculine and feminine) over the oblique cases. It is unclear whether such a hierarchy could be motivated in any independent way. The systematic syncretism of nominative and accusative in the neuter cannot be stated in terms of deletion of unmarked features, if neuter is an unmarked category as just assumed. An alternative is to link the expression of accusative to an abstract anim[ate] gender feature present only in noun phrases with masculine or feminine gender. I develop this line of thinking in §3 below. Weak nouns syncretize all non-nominative cases in the singular only. The fact that the singular is the unmarked value of number means that this cannot be stated in terms of a markedness cooccurrence restriction either (furthermore, it is unclear whether weak nouns are marked compared to strong ones).

16 2.5 Incremental weakening of the nominative-accusative distinction Classical Ovdalian¨ increases the number of systematic syncretisms slightly, compared to Old Swedish. The syncretism of nominative and accusative spreads through strong masculine and feminine singulars, and to the definite forms of the feminine. (34) Classical Ovdalian¨ ‘horse’, strong m ‘bridge’, strong f M indef def indef def sg N sg N est estn brune˛ A A bru D este estem D brun pl N ester estär pl N bruer bruär A esta esta˛ A D estum estum(e) D bruum

Thus, the only places in which the strong nominal paradigm distinguished nominative from accusative in Classical Ovdalian¨ were in the masculine plural (definite and indefinite). In the weak paradigm, illustrated below, nominative con- tinued to be distinguished from accusative in the singular of the masculine and feminine. (35) Classical Ovdalian¨ ‘bull’ m, wk ‘fly’ f, wk indef def indef def sg N ukse uksn sg N flugo flugo˛ A uksan A flugu˛ uksa flugu D uksam D flugun pl N ukser uksär pl N flugur A uksa uksa˛ A D uksum D flugum At this point, the distinction rests entirely on a diverse set of phonological features in weak syllables, as summarized in the following table. (36) Classical Ovdalian¨ Nom Acc str/wk m pl indef ester; ukser esta; uksa str/wk m pl def estär; uksär esta˛; uksa˛ wkmsgindef ukse uksa wkmsgdef uksn uksan wk f sg indef flugo, kulla flugu, kullu wkfsgdef flugo˛, kulla˛ flugu˛, kullu˛

Traditional Ovdalian¨ has completely neutralized nominative and accusative in the common noun paradigms. This has been concluded independently by, for example, Helgander (2005) and Garbacz and Johannessen (this volume). I provide additional evidence for this conclusion in section 3.1.

17 Levander noted that the last vestige of the distinction between definite and indefinite in the dative plural, in the strong masculine paradigm as seen in (34), was already rare in his 1909 investigation. The systematic syncretism can again be described as an Impoverishment rule. (37) def → ∅ / Dat,pl This is a typical Impoverishment rule, in eliminating a marked feature in the presence of some other marked features, but as with some of the other syncretisms discussed above, this statement of the facts runs afoul of my interpretation of the Feature Ranking Principle in (32), since definiteness should be hierarchically higher than plurality, given that definite articles are systematically structurally higher than plural markers. Since the definite plural suffix is usually -um, this generalization could equally easily be captured by positing an entry for -um which is underspecified for gender; no Impoverishment rule is needed. In the next subsection, I discuss a structural change from Old Norse to Classical Ovdalian¨ in the nominal suffixes which I argue is connected to this development.

2.6 Definite suffix fusion In Old Norse, the definite article was optionally encliticized to the noun, developing into a definite suffix in various daughter languages (e.g. Faarlund 2009). The definite marker can almost always still be neatly separated from the case- marked stem, as can be seen in the Old Swedish paradigms presented in the previous subsection. The only modification of the case-marked stem needed is generally describable in terms of deletion of final /r/, and occasionally of a final /a/, before the definite marker. I discuss an exception in n. 20 below. The inflection on the definite suffix inflects according to the ‘strong’ adjectival paradigm, generally showing the forms seen in adjectives ending in /n/, with an exception in the dative plural (where -inum would be the expected form, but the definite suffix is the uninflected form -in). Thus, the Old Norse nominal paradigms can be parsed into four parts: [i] the noun root, which is inherently specified for declension class and gender, [ii] a suffix which varies with declension class, gender, number, and case, [iii] the base of the definite marker, -n- with phonologically conditioned allomorph -in, and [iv] the adjectival agreement ending, showing gender, number, and case but not declension class. An indefinite noun only shows one suffix, but the one suffix carries information about case, number, and gender. This can be represented in a tree diagram as follows: N is dominated by a Cl[assifier] node, which carries information about whether the noun is singular or plural but also agrees in case (indicated by [Agr:K]) and in gender (indicated by [Agr:φ]); this node in turn is dominated by a case node K, which carries information about what case the noun phrase is but also agrees in number and gender (indicated by [Agr:φ]). When the two are adjacent, one suffix from the appropriate nominal declension series spells out both heads. Because the affix is adjacent to the noun root, it may be sensitive to declension class; thus thus a different stem such as siþ ‘custom’ can take a

18 diﬀerent form of the masculine plural nominative suﬃx, -ir.16 (38) K[Agr:φ]

Cl[Agr:φ,K]

hest -ar ‘horse’ m.pl.nom

In definite noun phrases, K has another dependent, D, which spells out as -n; this separates Cl from K, with the result that K spells out as a separate morpheme, from the adjectival series. Syntactic heads are typically assumed to be able to support at least two dependents, a complement and a specifier; in these terms, D is the complement of K, and Cl is the specifier of K (Cl may have moved from the complement position of D, perhaps motivated by the agreement probes). (39) K[Agr:φ]

Cl[Agr:φ,K] D

hest -ar -n -ir ‘horse’ m.pl.nom def m.pl.nom

The agreement probes on Cl and K ensure that they spell out the same set of features, except that only the morpheme which is adjacent to N can show allomorphy for declension class or for phonological properties of the nominal root. The alternation between /n/ and /in/ is determined by the phonology of the whole (/n/ if a vowel immediately precedes or follows), which can be assumed to reflect a late phonological rule. The neuter singular agreement marker is -t in the nominative and accusative case, and so the definite suffix would decompositionally be -in-t, but surfaces as -it (-iD in some varieties of Old Norse). This can be effected by a phonological rule, but it is possible that this suffix may have been reanalyzed at some point as a portmanteau consisting of D plus K, as shown in (40b).

16In the trees, straight lines indicate syntactic dependencies, while squiggly lines indicate exponence, assuming late spell-out, as in Distributed Morphology, or more speciﬁcally the implementation in Bye and Svenonius (2012).

19 (40) a. K[Agr:φ] b. K[Agr:φ]

Cl[Agr:K,φ]D Cl[Agr:K,φ]D

N N

skip ∅ -in -t skip ∅ -it ‘ship’ defn.nom/acc ‘ship’ n.def.nom/acc

Cl happens to be null in the neuter singular for this class of nouns, but some parts of the paradigm have case endings, even in the singular, which are expo- nents of Cl (the account remains essentially unchanged if Cl is replaced with two heads, Pl in plurals and Anim in masculine and feminine nouns; neither Pl nor Anim would be found in the neuter singular). In Classical Ovdalian,¨ the morphological paradigms show a different structure. There is no invariant definite affix. Instead, the definite affix and the case-gender-number morphology are fused. As a result there are never three suffixes, as there are in Old Norse. In some cases, there seem to be two suffixes, an inner one corresponding to Cl and an outer one corresponding to the fused D-K, but in other cases it seems there is only a single suffix, since even the part which signals definiteness can be sensitive to the declension class of the noun. First, consider the evidence that there are two suffixes. A neuter noun from one of the strong paradigms is given below with pitch accent marked on all disyllabic forms; an acute accent means ‘tone 1,’ and a grave accent means ‘tone 2.’ Tone 2 is the normal accent for disyllabic words, while tone 1 is like the accent on monosyllabic words. (41) ‘table’ n, str Classical Ovdalian¨ indef def sg N buord búordeD A D bùorde bùorde˛ pl N buord búorde˛ A D bùordum Some of the disyllabic words are tone 2, as expected: the dative ones. But the definite singular and plural nominative-accusative forms are disyllabic but tone 1. To make sense of this pattern, consider first the nominative-accusative singular. The definite suffix is -eD, added to a monosyllabic stem. The definite suffix is not counted for the purposes of tone. This pattern holds in many Scandinavian languages and shows that the definite suffix is less integrated into the word than the plural suffix (see e.g. Lahiri et al. 2005). The contrast can be seen in the examples in (42), from Nyström (1964).

(42) a. ... dar an ir ini fj˚a˛`se˛ dar wittern ir. there he is in barn.def.dat there winter.def is

20 ‘when he is in the barn when it’s winter’ b. kum˚a dier ini fj˚a˛´seD dar an ir dar, so werd dier come they in barn.def.drct there he is there so become they it gambler itjä! not old not ‘if they [rats] come into the barn when he [the cat] is there, they don’t grow old!’ Similarly, the definite plural nominative-accusative suffix -e˛ is added to a monosyllabic stem, and is not counted for tone, so the form surfaces as tone 1. In the dative forms, the singular has a dative suffix, which changes the syllable structure and hence the tone is 2. This suggests that the definite suffix in the dative singular is a nasal autosegment added to an already disyllabic base. If there were a single dative singular definite suffix -e˛, then it would give rise to tone 1, just like the definite plural nominative-accusative suffix. The minimal pair búorde˛–bùorde˛ is thus clearly understandable if Classi- cal Ovdalian¨ involves two nominal suffixes. Now consider a masculine noun. Here the definite singular nominative-accusative marker is a syllabic /n/, and is outside the domain of tone assignment; all other forms are disyllabic in the indefinite, and all other forms are tone 2. (43) ‘farm’, strong m M indef def sg N gard gárdn A " D gàrde gàrdem pl N gàrder gàrdär A gàrda gàrda˛ D gàrdum gàrdum

Again, the system is distinguishing plural suffixes, which consistently give tone 2, from definite suffixes, which do not. This is consistent with the plural suffix lexicalizing Cl, and the domain of tone being ClP; the definite suffix lexicalized the higher D-K projections, which are outside the domain for tone assignment. We can see from the indefinite singular dative form that case suffixes are also counted, suggesting that the dative suffix lexicalizes Cl, as in Old Norse. Now, consider whether two suffixes can be parsed in the definite forms. Clearly, the dative singular definite suffix could be /m/. The accusative plural definite suffix could be a nasal autosegment, just as in the neuter dative singular. And the nominative plural definite suffix could be a [low] feature which docks onto the vowel, causing /e/ to change to /ä/. If this is right, then the definite suffix is always added to the indefinite form, but the collection of definite suffixes, sensitive to the case and number of their host and sometimes autosegmental, is quite complex (and not transparently related to the system of adjectival inflection, as it was in Old Norse; see §3.6 for adjectival paradigms). Now consider a feminine noun from one of the strong declensions. Here, the definite nominative-accusative singular suffix /e˛/ and the definite dative suffix, a syllabic /n/, are outside the domain of tone assignment, resulting in tone 1, just as in the masculine paradigm, and the plural suffixes are inside that domain,

21 and give rise to tone 2, again like in the masculine paradigm. (44) ‘shed’, strong f indef def sg N búDe˛ A buD D bún pl N " bùDer bùDär A D bùDum If we can assume that a [low] autosegment can change the /e/ in the last syllable to /ä/, then this paradigm, too, could be analyzed in terms of concatenation of a Cl suffix and a definite suffix. However, there are complications. In the weak paradigm of some feminine nouns, for example, additional allomorphs would have to be posited to get the definite forms. Consider the alternation between indefinite plural nominative- accusative -er and definite plural nominative-accusative -ur. (45) ‘girl’, weak f indef def sg N kùlla kùlla˛ A kùllu˛ kùllu D kùllun pl N kùller kùllur A D kùllum The [low] autosegment posited for the strong feminine paradigm won’t work for the weak paradigm. If the affixes are sensitive to declension class, that suggests they are adjacent to the noun root, which would undermine the two-affix analysis just proposed. This could mean that in Classical Ovdalian,¨ a single portmanteau morpheme can spell out all three heads, as illustrated in (46a), while other parts of the paradigm involve combinations of suffixes, as illustrated in (46b).

(46) a. K b. K[Agr:φ]

D Cl[Agr:K,φ]D

Cl N

N kull -a [nasal] ‘girl’ f.sg.nom f.def.nom/acc kull -ur ‘girl’ f.pl.def.nom/acc

The single suﬃx is sensitive to the declension class of N and must therefore

22 be strictly adjacent to it.17 The Classical Ovdalian¨ suffixes are difficult to parse, which invites using this portmanteaux analysis for more of them. This suggests the spread of the portmanteau pattern in (46)a. Certain suffixes, like the definite singular masculine nominative-accusative /n/, could be lexically specified for tone. There are many languages in which roots seem to prefer to bear at most one inflectional suffix. For example, English verbs have at most one suffix, expressing tense (as in walked), agreement (as in walks), or aspect (as in walking). If there is eventually one suffix throughout the Ovdalian¨ paradigm, a structure like the one in (46)a would be consistent with the facts, but would leave the restriction on one affix per stem an unexplained stipulation. In fact, since several of the suffixes are underspecified, it becomes difficult to prevent them from cooccurring. What is to prevent a definite suffix like masculine singular Direct case -n from lexicalizing D, appearing outside a plural suffix like -er, lexicalizing Cl: *kall-er-n? Technically, such combinations can be prevented, but only by more fully specifying the lexical entries of the morphemes. If we were to adopt a one-suffix analysis throughout the paradigm, this would be more consistent with the following structure:

(47) Cl/D/K

kull -ur ‘girl’ f.pl.def.nom

Here, a single inflectional head bears gender, number, definiteness, and case features, and is spelled out by a single morpheme. Given that multiple heads are motivated in Old Norse, this proposal would require that head-bundling be parametrized (as argued in Bobaljik and Jonas 1996, Pylkkänen 2008). There is no evidence apart from the morphology that the Ovdalian¨ noun phrase contains less structure than the Old Swedish ones. An alternative analysis, which preserves parallelism in structure, would be the following:

17The same can be observed in at least some dialects of Norwegian; for example, there are speakers for whom barna is the definite form of neuter plural barn ‘child’, but husene is the definite form of neuter plural hus ‘house’; there is no phonological generalization describing the distribution, and thus both -a and -ene are neuter plural definite suffixes, distinguished only by declension class, and so must spell out a node which is adjacent to the root N. See Svenonius (to appear).

23 (48) K

Cl[Agr:φ,D,K]

kull -ur ‘girl’ f.pl.def.nom

The idea here is that K and D are systematically null in Ovdalian,¨ but the agreement probe on Cl copies D features as well as K and φ. The fact that K and D are always null is a stipulation, but may be easier to justify formally than the situation with the first hypothesis, which required a large class of portmanteaux and some kind of conspiracy to prevent bimorphemic endings. In any case, if there is a single suffix, then the loss of the definiteness distinction in the dative plural can now be described in terms of the tendency for morphemes to express fewer features; dative, plural, and definite are all marked features and so there is a pressure to express fewer of them. In Old Norse, only -n expresses definite, and it does not express any other features (eventually -it also expresses definiteness along with K). This reanalysis is also supported by phonological facts. In Old Norse, the phonological boundary between the case and number-marked stem and the definite marker is phonologically relatively salient. The boundary is significant for phonological rules such as umlaut; a /Y/ in a Cl-suffix induces umlaut in the stem, but a /Y/ in a K suffix does not (though as we saw in Old Swedish, regres- sive deletion of /r/ and /a/ did occur across this boundary). Thus Cl is inside a cycle of morphophonology that excludes D. In Classical Ovdalian,¨ however, there are indications that the definite marker is a more phonologically integrated suffix, indications which point in exactly the opposite direction from the tonal evidence. Suffixes which begin with front vowels trigger palatalization of velar codas in the root, as seen in (49) (using the orthographic hdji for /dZ/ and htji for /tS/; I take the /e/ of the nominative definite singular and the nominative indefinite plural to be epenthetic here). (49) Paradigms for Classical Ovdalian¨ ‘wolf’ and ‘grub, maggot’ M indef def M indef def sg N wargen sg N makken warg makk A wardjin A mattjin D wardje wardjem D mattje mattjem pl N warger wargär pl N makker makkär A warga warga˛ A makka makka˛ D wargum D makkum Notice in particular that there is no palatalization in the accusative indefinite singular, but there is in the accusative definite singular, the only featural dif-

24 ference being definiteness; so definiteness is phonologically integrated with the stem in a way that is characteristic of a suffix, not of a clitic. The accent, as indicated by Levander (1909), is nevertheless tone 1: wárdjin (compare dative singular definite wàrdjem with tone 2, as expected). If the palatalization shows that -in is inside the smallest phonological domain, despite its tone neutrality, then it must also be lexically specified with metrical structure to prevent the intonational pitch accent of the stem from including it.18 Additional evidence that the suffixal material in Classical Ovdalian¨ is more well-integrated than the definite parts of the suffixal complex in Old Norse is that there are irregular stems which are sensitive to the features in the suffix; for example tjyr ‘cow’ has a stem-final /r/ in the singular but not the plural forms: nominative/accusative plural tjyner, dative plural tjym (Levander 1909:35) rather than*tjyrar and *tjyrum.19 In Old Norse, only the material up to Cl can be involved in irregularity, not the material beyond it. These considerations support the hypothesis in (48) over various alternatives.20 The fusion of the suffixes is an important difference between the Old Norse nominal inflectional system and the Classical Ovdalian¨ one: the Old Norse noun has up to three suffixes, only one of which is strictly speaking part of the nominal paradigm. The second suffix out is the largely invariant -(i)n, with a phonologically motivated alternation but very little other allomorphy, and is linked to the paradigm for adjectives and participles, and the third part of the noun’s morphology is identical to adjectival agreement. Thus, for example, a systematic syncretism in the weak paradigms of accusative, dative, and genitive singular need not have any effect on the definite suffix, which follows the adjectival paradigm. Furthermore, the definite forms are always distinct from the indefinite forms for the simple reason that the definite suffix is overt and is added to the indefinite form. In contrast, the Classical Ovdalian¨ nominal suffixes, at least for parts of the paradigm, appear to be fully fused, with no morpheme boundary distinguishing the definite part from the inner part. This could be expected to accelerate feature neutralization, since it means that a single suffix is bearing more of a featural burden, and without reinforcement from other paradigms.

3 From Classical Ovdalian¨ to Traditional Ovdalian¨

Traditional Ovdalian¨ has completely neutralized nominative and accusative in the common noun paradigm, as indicated in the table. I provide evidence for this in section 3.1. 18I am assuming an analysis of accent like the one proposed in Morén-Duolljá(2013). Thanks to Patrik Bye for discussion. 19Steensland (2006:109) notes an alternate dative plural form tjynum, which would be a regularization of the plural paradigm. 20Delsing (2002) notes one place in Old Swedish which might suggest a similar development. In the dative and genitive singular of ¯o-stem feminine nouns, the inner suffix is unexpectedly missing before the definite suffix. That is, one apparenly finds s¯ol-inni rather than the expected *s¯ol-u-nni (dative), and s¯ol-innar rather than *s¯ol-a(r)-(i)nnar (genitive).

25 (50) Classical Ovdalian¨ Traditional Ovdalian¨ Nom Acc Nom/Acc = Drct str/wkmplindef ester;ukser esta;uksa ester;ukser str/wk m pl def estär; uksär esta˛; uksa˛ ester; ukser wkmsgindef ukse uksa uksa wkmsgdef uksn uksan uksn(oruksan) wk f sg indef flugo, kulla flugu, kullu flugu, kulla wkfsgdef flugo˛, kulla˛ flugu˛, kullu˛ flugu˛, kulla˛ Since the language must have accusative case in order to license it on pronouns, and since the distribution of full noun phrases is the same as that of pronouns, it seems that accusative case exists in the language, it is simply not expressed on common nouns. This can be described in terms of an Impoverishment rule, taking Cl to be the locus of gender and hence of the anim feature posited to be present on masculine and feminine nouns but not neuter ones. The rule only applies to lexical nouns (N), not to pronouns. (51) Acc → ∅ / Cl, N Note that this rule does not seem readily compliant with the Feature Ranking Principle in (32), as it would require N to be ranked higher than Accusative. According to the analysis in (48), K in Classical Ovdalian¨ is never overt. What is overt is agreement in case, on the Cl head (and certain adnominal elements). Agreement probes do not always copy case features; that is why [Agr: K] and [Agr: φ] had to be distinguished. It seems plausible, then, that the loss of accusative on full nouns is not due to Impoverishment at all, but rather to a change in the nature of the agreement probe on Cl. Suppose that Cl in Traditional Ovdalian¨ probes for φ and for dative, but not for other cases. Then Cl will never pick up accusative features. This resolves the tension with the Feature Ranking Principle, since there is no longer any syncretism to be described: Cl simply doesn’t have accusative features. The systematic syncretism of nominative and accusative in the neuter in Old Norse, noted in §2.4, could be handled similarly, if neuter nouns lacked a probe which masculine and feminine nouns had. The same approach might also reconcile the underspecification of nonnominative case in weak singular noun paradigms also noted in §2.4. Pronouns, arguably, have a different structure from common nouns; Cl is the locus of plurality in common nouns, but plural pronouns are different (first person plural is not a plurality of speakers, nor is second person plural necessar- ily a plurality of addressees). So pronouns can have their own [Agr: K], or even spell out K directly, without that affecting the common noun paradigm. The Silverstein hierarchy, which states that pronouns are more likely that common nouns to express accusative case, suggests that this is a common situation. In the next section I present evidence for the neutralization.

3.1 Accusative Levander (1928:128) notes that the distinction between nominative and accusative in full nouns was wavering in the speech of informants living in the larger towns in Alvdalen¨ already in the 1920’s, and lost in neighboring districts.

26 He furthermore reports the absence of the distinction among the young even in the smaller northern village of Asen,˚ specifically in an informant born in 1909 and his peers—noting that the same youth’s 10 elder siblings retain the distinction (ibid, p. 316 n. 45). He give the examples in (52), where (52b) reflects the loss of the accusative (ibid, p. 128).21 (52) a. Sir du ettu˛? see you hat.def.acc ‘Do you see the hat?’ b. Sir du etta˛? see you hat.def ‘Do you see the hat?’ When the nominative and accusative collapse into a direct case form, the form which takes over is sometimes the historical nominative and sometimes the historical accusative, hence the observation of an historically accusative form is not evidence that a distinction is being made. Still, it is convenient to speak of the accusative as being what was lost in diachronic terms.22 Helgander (1996) reviews evidence suggesting that the distinction was first lost in indefinite forms among speakers born at the end of the nineteenth century and then in definite forms, with speakers born in the 1930’s showing no nominative-accusative distinction on full nouns. The loss of the distinction is illustrated below in a pair of examples from Nyström (1964:15, 22f), an interview with a speaker from the northern village of Karlsarvet, born in 1891. The orthography has been standardized (with the assistance of Piotr Garbacz), except for the endings on the full nouns (a hesitation marker is marked with angle brackets). (53) a. ... ses eD wart graise jär, so add – dier add graistjitte since it been pigs here so had they had pig.pens ‘. . . since there were pigs here, they had – they had pig pens’ b. . . . dier so ˚av˚a graiser, so ˚av˚a dier grais- hei tjittär aute they as have pigs so have they pig eh pens.def outside ‘. . . those who have pigs, they have [the] pig, eh, pens outside’ Grais ‘pig’ and tjitte ‘pen’ are masculine nouns belonging to different declensions. In Classical Ovdalian,¨ both would have had nominative plural -er in the indefinite, and -är in the definite, in Karlsarvet (Levander 1909:11, n. b). Both would have had accusative plurals in -a in the indefinite and in -a˛ in the definite. We expect nominative in the existential construction ‘there were pigs,’ and the loss of final /r/ in this phonological context is common, so the first form graise[r] in (53a) is as expected. However, we expect accusative after ‘have,’ so the form graiser in (53b) suggests a loss of the nominative–accusative distinction. Similarly, the appearance of historically nominative forms graistjitte[r] and graistjittär in accusative contexts in (53a–b) suggests the loss of the distinction.

21Specifying only unambiguously expressed features in the gloss, in order to avoid implying more distinctions than are actively being made in the variety being described. 22Steensland (2006:12) observes that the historically accusative form of singulars often becomes the direct case form for masculine nouns in -e and a large class of feminine stems ending in vowels (a class with an open ﬁrst syllable, hence not including etta ‘hat’ in (52), which goes the other way, generalizing the historical nominative).

27 Today the nominative-accusative distinction on full nouns can be considered archaic at best, even among the most conservative speakers. This is implic- itly recognized in Steensland’s (2006) dictionary, which provides dative forms but not distinct accusative forms. Knowledgeable elderly speakers know the accusative forms and sometimes regard them as correct in accusative contexts, and provide them in elicitation, but I have not been able to observe a nominative- accusative distinction being made in spontaneous speech. More commonly, speakers neutralize the nominative and accusative, as in the (elicited) example below. (54) Sir du ester mainer? see you horses my.pl ‘Do you see my horses?’ Here, the accusative form would have been esta˛ maina in Classical Ovdalian.¨ Below is a minimal pair of elicited sentences showing the masculine noun lapp ‘patch’ in both nominative and accusative contexts, again syncretized (the dative on oll ermum ‘all the sleeves’ in (55) correlates with the static description, while the corresponding accusative (i.e. ‘direct’ case) in (56) correlates with the dynamicity of patches moving onto sleeves). (55) ED ir lapper o˛ oll ermum. it is patches on all the.sleeves.dat ‘There are patches on all the sleeves’ (56) Ig sit og sömer lapper o˛ ermer. I sit and sew patches on sleeves ‘I am sitting down sewing patches on the sleeves’ As this is a strong masculine noun, the accusative form would have been lappa (indefinite) or lappa˛ (definite) in Classical Ovdalian.¨ Some examples from interviews conducted in 2007 by Lars Steensland (personal communication) follow. (57) a. ... og best lärerer amm wi˛D aft and best teachers have we had ‘... and we have had the best teachers’ b. ... og dar warum wi˛D um kwelder og and there were.1pl we about evenings also ‘... and we were there in the evenings too’ The same informant lacks dative in the following example where it would have been expected after this sense of the preposition min ‘with’ (though min has a variety of senses, and some of them regularly take the direct case, historically accusative). Coordination is a common context for the lack of expected dative, for some reason. In the example in (58), the feminine nouns dukka ‘doll’ and tjyr ‘cow’ are not expected to distinguish nominative from accusative, but the masculine noun est ‘horse’ is. (58) daruppi belld ig byddja ... og min dukkur og träester og there.up could I live and with dolls and wood.horses and gr˚a˛tjyner og ollt eD-dar spruce.cows and all that-there

28 ‘I could live up there ...and with dolls and wooden horses and spruce- cone cows and all that’23 Finally, some examples from KK. First, there are examples in KK where the nominative-accusative distinction appears to be made. For example, the strong masculine noun kripp ‘child’ appears in the plural forms kripper and krippa, as seen below. I gloss this form with ‘acc’ in parentheses since it is at issue whether this form is really consistently used in accusative contexts. (Note also that nasalization is only sporadically marked in KK, so the example in (60) might actually be krippa˛, i.e. the definite-marked accusative plural.) (59) Kripper add kurad ijuop sig sos smo˛m˚ajser i ie trong, swart children had huddled together self like small.mice in a tight black wr˚a i gamtstugun. corner in the.old.house ‘The children had huddled together like little mice in a cramped dark corner of the old house’ (60) An add remt ad en, og spuort on ur kringt an add he had bellowed to her.dat and asked her.acc how often he had lesid upp Salomos uordspr˚ak fer en, fer at o˛ uld lä’ read up Solomon’s words for her.dat for that she should learn sig, ur o˛ uld antir krippa. self how she should manage children.(acc) ‘He had bellowed to her, and asked her how often he had read aloud Solomon’s words for her, so that she would learn how to govern the children’ In the same text, kripper also appears in contexts which should be accusative. (61) Twe kellinger, og nog kripper, add Alwar nest sig, og oller fing two women and some children had Alwar by self and all got minsann dsjä rett fe’ sig. at.least do right for self ‘Two women, and some children, Alwar had with him, and everyone was to account for themselves’ (62) Nu ir fr˚aga, um itt mickel auti diem da 200 man, so now is the.question if not many out.of those there 200 man as uld fya, add keling og kripper iem, so add styerr should follow had woman and children home who had greater anliening war aingsliger eld wen prestfrune˛ adde. reason be anxious than what the.priest.wife had ‘Now the question is, whether many of those 200 men who were going to follow, didn’t have a wife and children at home, who had more cause to be anxious than the priest’s wife did’ The examples above show coordination, which occasionally (though not always) interferes with dative case (in addition, the first involves displacement, which is noted by Sandøy (2000) to inhibit the assignment of dative in Norwegian

23According to Lars Steensland (personal communication), spruce cones are used to represent cows in children’s play, while pine cones represent sheep.

29 dialects with dative case). However, the absence of a nominative-accusative distinction goes beyond such circumstances and is far more common in KK, even more common than strict observance. Two examples are given below. (63) So dier lat kripper f˚ar˚a, so they let children travel ‘So they let the children travel, ...’ (64) Dar kasted o˛ dier dar sju dieler attrywyr yegeraxeln there threw she those there seven pieces back.over the.right.shoulder (wister um edd a’ we’ fr˚agan um ie kulla). left if it had been the.question of a girl ‘There, she threw those seven pieces back over her right shoulder (left if it [the exorcism] had concerned a girl)’ The examples above have all involved the extension of an historically nominative case into accusative contexts, but the substitution of forms goes both ways. An example of an historically accusative form in a nominative context is seen below. The example contains both the historically nominative and the historically accusative form. (65) Jälln war uppdsildrad min dsilderstickur, so da’ byenn the.platform was up.trapped with trap.sticks so when the.bear kam a’ diem, so räsed jällan nid o˛ an, og klemd came to them so fell the.platform.(acc) down on him and crushed ijel a˛n. to.death him ‘The platform was booby-trapped with spikes, so that when the bear came toward them, the platform fell down on him and crushed him to death.’ Similarly, in the example below, maru ‘(the) nightmare’ (a supernatural being) is the subject, and should be nominative (mara, a weak feminine noun, in older forms of the language); in the second example, the same noun is an object (or subject of an ECM construction), and should be accusative. Yet it is typical of KK that the forms are not distinct.24 (66) E’ wa so, at da maru uld rus fro˛ gardem, dar it was so that when nightmare should travel from the.farm.dat there o˛ add bitid offred, so bruked o˛ t˚ago og raid o˛ ien est she had bitten the.victim so used she take and ride on a horse ‘It was so, that when the nightmare was ready to leave the farm where she had bitten a victim, she would take a horse and ride it’ (67) o˛ add la’t maru f˚a sos o˛ wild i gardem, og nu she had let nightmare get as she wanted in the.farm.dat and now s˚ag o˛ ur ed add gajd. saw she how it had gone ‘She had let the nightmare have her way on the farm and now she saw

24As noted by Lars Steensland (see n. 22 above), this extension of the accusative/dative is typical of short-stemmed weak feminine nouns, while long-stemmed weak feminine nouns such as tjy¨orta ‘church’ tend to generalize the nominative.

30 how it had gone’ Thus, although accusative forms are sometimes found in accusative contexts in Traditional Ovdalian,¨ the system as a whole does not make this distinction, for full noun phrases. On the other hand, the syntax makes a distinction among nominative and accusative case contexts, since ﬁrst and second person and plural pronouns systematically continue to distinguish the ﬁnite subject position from other positions, with paradigms essentially like those described in section 2.2. The three-way distinction for the feminine pronoun made in Classical Ovdalian¨ appears to be collapsing, however, see Garbacz and Johannessen (this volume), meaning that pronouns generally seem to make a two-case distinction, setting aside the possessive forms. In the next section I document the persistence of the dative case in conservative varieties of Traditional Ovdalian.¨

3.2 Dative Dative case appears fairly consistently on the objects of certain prepositions. As is typical of Indo-European languages with case, there are prepositions which govern accusative, prepositions which govern dative, and prepositions which alternate, with static descriptions requiring dative and dynamic ones requiring accusative, in the manner familiar from languages like German. This system is largely preserved in Ovdalian,¨ illustrated here with some elicited examples.25 (68) a. Brindn fuor upo˛ bokkan. the.moose went up.on the.hill ‘The moose went up the hill’ b. Brindn stand kwer upo˛ bokkam. the.moose stands still up.on the.hill.dat ‘The moose is standing still up on the hill’ (69) a. Brindn k˚ait inunder brune˛. the.moose ran in.under the.bridge ‘The moose ran under the bridge’ b. Brindn stand under bru’nn. the.moose ran in.under the.bridge.dat ‘The moose is standing under the bridge’ (70) a. An tuog sig ini baureD. he took self in the.storehouse ‘He made his way into the storehouse’ b. og s˚amneD ini baure˛. and fell.asleep in the.storehouse.dat ‘and fell asleep in the storehouse’ Examples of this can be heard in spontaneous speech. Below is a pair of examples from Nystr¨om (1964), of a particularly conservative variety,26 showing

25Classical Ovdalian¨ distinguishes nominative bottjin ‘the hill’ from the accusative form bokkan, but since this distinction is not consistently made in the variety being described, I do not gloss bokkan here as accusative. 26The speaker was born in 1891, well before the stipulated boundary between ‘Classi- cal Ovdalian’¨ and ‘Traditional Ovdalian,’¨ but already showing at least partial loss of the

31 the feminine noun rotta in two forms, nominative/accusative plural (rottur) and dative plural (rottum; broad transcription due to Piotr Garbacz). (71) An ir it ... sakt duktin jag og t˚ag˚a m˚aiser, an ar fel he is not truly talented hunt and get mice he has you.know taiD boD is-jär sturrottur og! taken both these-here big.rats also ‘He isn’t . . . truly is good at hunting and catching mice, of course he has also caught these big rats’ (72) Sleppt aut diem-dar rottum! released out those-there rats.dat ‘[They] let those rats out!’ More examples are given below from KK, preserving the (non-standard) orthography employed by Hjalmar Larsson in that work, except to mark the unambiguously dative-marked noun in boldface. First, (73) illustrates a masculine definite singular (gard ‘yard, farm,’ compounded with tsyerts(a) ‘church’) and (74) shows two masculine plurals (munn ‘mouth’ and est ‘horse’).27 (73) Viktugestad war, um an add no˛d fro˛ tsyertsgardem, most.important was if he had something from the.churchyard.dat non spajk fro˛ ie laiktsist, eld no˛d bien og lited muld. some nail from a coffin or some bone and a.little soil ‘The most important was if he had something from the churchyard, a nail from a coffin or some bone and a little soil’ (74) E’ kam raidendes noger förridderer, so ed froded yr it came riding some advance.riders so it frothed out.of munnum ad estum. mouths.dat to horses.dat ‘Some scouts came riding so that their horses’ mouths were frothing’ Example (75) is a feminine definite singular (stugu ‘building, house’), and (76) is a feminine plural (p˚asker ‘Easter’ is a plurale tantum). (75) Ad stugun ärd og iet wiselt fjo˛s og iet a’byggt stoll. to the.house.dat belonged also a shabby barn and a to.built stall ‘Along with the house was a dilapidated barn and an added-on stable’ (76) men fö’ säkerits skull, so mo˛led an um ed ad p˚askum but for security’s sake so painted he about it to Easter.pl.dat wert ˚ar. every year ‘but to be on the safe side he repainted it at Easter every year’ Finally, an example of a definite singular neuter noun (buord ‘table’), and a plural neuter noun (oga ‘eye’).

nominative–accusative distinction, as discussed above 27Another note on glossing conventions: As I detail below, this variety of Ovdalian¨ does not overtly mark deﬁniteness distinctions in the plural; thus, following the conventions mentioned in n. 21, even plurals which are clearly understood as deﬁnite are not glossed as such.

32 (77) Kunundsin tuog plass nest buorde min knektrullur. the.king took place by the.table.dat with soldier.lists ‘The king took a seat at the table with the list of soldiers’ (78) An wa l˚ag o˛ mo˛led dar an spr˚aked, og an bögles aldri i he was low on the.voice there he spoke and he stared never in ogum o˛ diem an spr˚aked min. eyes.dat on them he spoke with ‘His voice was low when he spoke, and he never looked right in the eyes of the person he was speaking with’ The patterns above are robust in KK, though there is some degree of variation. Generally, plurals and definite singulars have unambiguous dative case marking when they immediately follow a dative-governing preposition. Most failures of dative case marking involve some complication, for example in the coordinate structure in (79), only the first conjunct is dative-marked (the dative plural of finkall ‘nobleman’ would be finkallum). (79) E’ war iet buord autsett fe’ kunundsem og finkaller. it was a table set.out for the.king.dat and noblemen ‘There was a table set out for the king and the noblemen’ A dative form appears on the possessor in possessor constructions, as shown in the following example.28 (80) Marit sett igo˛ng, og sambled ijuop sju dieler fro˛ Marit sat in.walk and collected together seven parts from Lihl-Spajtsem es krupp. The.Little-Nail.dat poss body ‘Marit got busy and gathered together seven parts from Lihl-Spajk’s body’ In such constructions, dative case often fails to appear on the head noun, even in environments which otherwise robustly require dative. In the following example, the head noun klauter is unambiguously in the direct case (dative plural is klautum), despite heading the complement of fro˛ ‘from,’ which systematically takes dative. (81) Rästn tuog Marit fro˛ Lihl-Spajtsem es klauter the.rest took Marit from The.Little-Nail.dat poss clothes ‘The rest, Marit took from Lihl-Spajk’s clothing’ In a few cases dative can be seen on direct objects of certain dative-governing verbs, such as j˚apa ‘help.’ (82) Ennes metuoder add itt j˚apt nogum juottedags, og ka[m] her methods had not helped anyone.dat hither.days and came it ta dsjäro ed eld. not to do it either

28As elsewhere, I preserve the original orthography from KK ; Lars Steensland points out that current practice would favor writing the es together with the name, reﬂecting the per- ception that they form a single word.

33 ‘Her methods hadn’t helped anyone up to then, and weren’t going do so, either’29 Some investigators, for example Garbacz and Johannessen (this volume), have observed the instability of the dative among contemporary speakers. The pattern described here should be considered a conservative variety of Traditional Ovdalian.¨

3.3 Neutralization of definiteness distinctions in the plural A striking development in Traditional Ovdalian¨ is the neutralization of definite distinctions in the plural. In section 2.5, I pointed out the neutralization in Classical Ovdalian¨ of definiteness in the dative plural, describing this in terms of Impoverishment in (37). In Traditional Ovdalian,¨ this pattern of syncretism is more general, affecting the Direct case plurals of masculine and feminine nouns. The following tables represent paradigms which I believe to be representative of six classes of nouns which had systematically different behavior in Classical Ovdalian.¨ There is one caveat: dative might be more frequently marked in the indefinite singular forms than I have been able to detect; see the discussion of apocope and of pitch accent possibly marking dative case in §2.6. Setting that possibility aside, there seems to be widespread neutralization of the dative-direct distinction in the indefinite singular, just as in the dative-preserving dialects of Norwegian studied by, e.g. Eythórsson et al. (2012). In each pair, the noun on the left is from a strong declension class and the one on the right is weak; the examples in (83) are masculine, those in (84) are feminine, and those in (85) are neuter. Boxes indicate syncretisms, and Drct stands for ‘direct case’ the nominative–accusative. Spellings are adapted from KK.30 (83) kall ‘man’ (str. m.) jälle ‘platform’ (wk. m.) M indef def M indef def sg Drct kalln sg Drct jälln kall jälle Dat kallem Dat jällam pl Drct kaller pl Drct jäller Dat kallum Dat jällum (84) asp ‘asp’ (str. f.) tjyörtja ‘church’ (wk. f.) F indef def F indef def sg Drct aspe˛ sg Drct tjyörtja˛ asp tjyörtja Dat aspin Dat tsjörtjun pl Drct asper pl Drct tjyörtjur Dat aspum Dat tjyörtjum (85) tak ‘roof’ (str. n.) ära ‘ear’ (wk. n.)

29In the original text (KK p. 41), the word kam ‘came’ is spelled kan, which is a form of the modal ‘can,’ but doesn’t suggest a plausible parse; I assume following a suggestion of Lars Steensland that kan in the text is a typographical error for kam. 30As already noted, the direct definite singular jälln alternates with jällan, so the form could be represented as jälln∼jällan.

34 N indef def N indef def sg Drct tatjeD sg Drct äraD tak ära Dat tatje˛ Dat ära˛ pl Drct tak tatji˛ pl Drct ärur Dat takum Dat ärum There are more noun classes than are displayed here, but these are broadly representative (for more examples see §3.5). The paradigms displayed here are an idealization, and the actual data collected shows much variation. The total number of suffixes involved is not very great, so it is not clear how much would be gained by stipulating an Impoverishment rule; the suffixes -er and -ur and any others could simply be underspecified, as I suggested in §2.5 for dative plural -um. The syncretism of the definite and indefinite plurals might somehow be a consequence of the fusion of the definite and plural suffixes. However, some other Mainland Scandinavian languages have also fused their plural and definite suffixes (Faarlund 2009), without the loss of definite distinctions in the plural (on the other hand, gender distinctions in the plural are often lost in those languages).

3.4 The indefinite singular dative Classical Ovdalian¨ had a dative form ienum of the masculine indefinite article, but this seems to be largely absent from spontaneous speech in Traditional Ovdalian,¨ where the direct form ien is substituted. Overt dative suffixes on indefinite singulars are frequently absent from KK, as in the following examples. (86) so bruked o˛ t˚ago og raid o˛ ien est, um e’ wa non i nerietn. so used she take and ride on a horse if it was any in the.nearby ‘she would take a horse and ride it, if there was one nearby’ (87) att dier so tuog and um djäster, uld werd bifriader fro˛ that they as took hand about guests should become absolved from skatt. tax ‘that those who took care of guests would be relieved from tax’ (88) o˛ uld old aused, ausfuotsed, estn og krytyre she should hold the.house the.house.people the.horse and the.creatures riener fro˛ ollt uotyg so Satan lat usir auti diem. clean from all vermin as Satan let ravage out.on them ‘She was supposed to keep the house, the people, the horse and the animals clean of all the vermin that Satan unleashed upon them’ In older forms of the language, strong masculine and neuter nouns like est, skatt, and u˛otyg would have a vocalic suffix in the dative singular; in KK the preposition fro˛ always takes dative case when the complement is plural or definite. Yet in these indefinites no dative morpheme is seen. Even the conservative text in Nyström (1964) contains an example which lacks the dative ending on an indefinite singular, e.g. the following example (ibid. pp. 37-38, transcribed in an adapted standard orthography from Nyström’s

35 narrow phonetic transcription). (89) ED m˚a˛tt d˚a˛ w˚ar nog so war nog flytend dier blätt it must then been something as was something liquid they dissolved noglund i nog liuotkräld ˚av noger somehow in some nasty.vessel of something ‘It [the fly poison] must have been some sort of liquid that they dissolved somehow in some nasty container of some kind’ Here the dative singular of the strong neuter kräld ‘vessel’ would have been krälde in Classical Ovdalian.¨ A phonological process of apocope, common in the region, may be the culprit. In many cases of apocope, the pitch accent distinctions on the affected word remain as if the deleted syllable were still present. As I showed above, dative singular indefinite case suffixes on strong noun stems add a second syllable and hence give rise to tone 2. This means that if the dative suffix is deleted by apocope, then in some cases it might be recoverable, in that the accent on the monosyllabic noun might be that of a tone 2 word, contrasting with the direct case tone 1.31 This possibility cannot be investigated in KK, because tone is not marked. I also found it difficult to investigate in field work, owing partly to the scarcity of relevant examples but also the difficulty, for one not adept in the language, of distinguishing tone 1 from tone 2 on monosyllabic words. In the narrow phonetic transcriptions of Gunnar Nyström, however, it is clear that a tonal contrast is made on monosyllabic words, and the monosyllabic words with tone 2 would in most cases have had a syllabic suffix in earlier forms of the language. The compound noun liuotkräld ‘nasty vessel’ in (89) is too long to show the effect, and I have not found any other indefinite datives in Nyström (1964) missing their dative suffixes. But there are several words in that interview which preserve tone 2 despite missing expected vocalic suffixes, for example the verbs m˚a`tt[e] ‘had to’ and blä`tt[a] ‘dissolved’ in (89), here with the missing vowels supplied and the tone marking added. The following transcriptions are regularized to the standard Ovdalian¨ orthography,32 except that the expected—but crucially missing—vocalic ending in each example is supplied in square brackets, and the tone from Nyström’s narrow transcription is added to the words in question. (90) a. so eD ar laiD nogu˛ slaik l`yöD[a] dait˚a˛ fj˚a˛sguove˛ um so it has lain some such lump there.on barn.floor.def.dat on me’nner milumaD mornings occasionally ‘so there has been a lump like that lying on the barn floor in the mornings, sometimes’ b. Ja wiD addum i˛e fòll[u] so tuogum boD sturrottur i og yes we had.1pl a.f trap as took also big.rats in too

31I am grateful to Lars Steensland for suggesting that I investigate this possibility. Thanks also to Patrik Bye for discussion. 32In this case by Piotr Garbacz. E.g. the original narrow transcription mà˛ne$ ‘mornings,’ an irregular masculine noun, is standardized to me’nner. The fact that the vowel˜ in the ending is [e$ ], i.e. /e/, in an accusative context, is typical of the simplification of the direct plural endings in Traditional Ovdalian:¨ Levander (1909:16) notes the nominative definite plural form mè:nnär, and implies an accusative indefinite plural form mè:nna, definite plural mè:nna˛.

36 ‘Yes, we had a trap that also captured big rats’ These are weak nouns; l`yöDa would be a nominative form, and fòllu would be accusative (or objective), so they do not prove that dative case can be distinguished from nondative by tone alone in the right context (namely, a strong noun which would have tone 1 in the nominative-accusative singular); but the general pattern of tone 2 preservation in what appear to be reduced forms suggests that it is definitely a possibility. In fact, a case marker which is realized on the surface only as a tone shift could easily be reanalyzed as an autosegment, for example a floating mora. It remains to be seen whether this is the best analysis for any actual Ovdalian¨ case alternations.

3.5 Comparative Paradigms To provide a sense of the general system, I present a representative assortment of nominal paradigms comparing the three-case Classical system to the newer two-case system. I present one each of masculine, feminine, neuter in each of the strong and weak declension classes. Again, it must be remarked that the indefinite dative singular may not be systematically absent from Traditional Ovdalian,¨ if it preserved in tone, which has not been adequately studied. (91) ‘man’ m, str Classical Ovdalian¨ Traditional Ovdalian¨ indef def indef def sg N sg Drct kalln kall kalln kall A Dat kallem D kalle kallem pl N kaller kallär pl Drct kaller A kalla kalla˛ Dat kallum D kallum (92) ‘bull’ m, wk Classical Ovdalian¨ Traditional Ovdalian¨ indef def indef def sg N ukse uksn sg Drct uksn uksa A uksan Dat uksam uksa D uksam pl N ukser uksär pl Drct ukser A uksa uksa˛ Dat uksum D uksum (93) ‘bridge’ f, str Classical Ovdalian¨ Traditional Ovdalian¨ indef def indef def sg N sg Drct brune˛ brune˛ bru A bru Dat brun D brun pl N pl Drct bruer bruer bruär A Dat bruum D bruum (94) ‘fly’f,wk

37 Classical Ovdalian¨ Traditional Ovdalian¨ indef def indef def sg N flugo flugo˛ sg Drct flugu˛ flugu A flugu˛ Dat flugun flugu D flugun pl N pl Drct flugur flugur A Dat flugum D flugum (95) ‘table’ n, str Classical Ovdalian¨ Traditional Ovdalian¨ indef def indef def sg N sg Drct buordeD buord buordeD buord A Dat buorde˛ D buorde buorde˛ pl N pl Drct buord buorde˛ buord buorde˛ A Dat buordum D buordum (96) ‘ear’n,wk Classical Ovdalian¨ Traditional Ovdalian¨ indef def N indef def sg N sg Drct äraD äraD ära A ära Dat ära˛ D ära˛ pl N pl Drct ärur ära äru˛ A Dat ärum D ärum

3.6 Adnominal modifiers Besides lexical nouns and pronouns, case is often also seen on various adnominal elements such as determiners, quantifiers, numerals, and adjectives. In Classical Ovdalian,¨ such elements provided rich case information, as seen in the paradigms for the indefinite determiner noger ‘some, any’ here, illustrated with the nouns kall ‘man,’ kulla ‘girl,’ and aus ‘house.’33 (97) Classical Ovdalian¨ M F N Sg Nom nogerkall nogu˛ kulla noD aus Acc non kall nog kullu noD aus Dat nogum kalle nog kullu nog ause Pl Nom nog kaller nog kullur nogu˛ aus Acc nog kalla nog kullur nogu˛ aus Dat nogum kallum nogum kullum nogum ausum These systems are considerably simplified in modern colloquial varieties, as seen in the following table.

33Additional inﬂected forms are used when not adnominal, i.e., when used as the only element in a noun phrase, for example feminine singular accusative noger(a), dative noger; masculine plural nominative nogrer, accusative noger(a).

38 (98) Traditional Ovdalian¨ M F N Sg Drct non kall nogu˛ kulla noD aus Dat non kall nogu˛ kulla noD aus Pl Drct nog kaller nog kullur nog(r)u˛ aus Dat nog(um) kallum nog(um) kullum nog(um) ausum As can be seen in the table, gender is distinguished in this system, but no gender distinguishes case in the singular. More and different distinctions are seen in freestanding pronouns, but adnominal determiners generally seem to have lost the dative singular form, and for many speakers also the dative plural. Levander (1909) gives case-inflected forms of adjectives for Classical Ovdalian.¨ (99) Classical Ovdalian¨ M F N Sg Nom stur stur sturt Acc sturan stur(a) sturt Dat sturum stur sturo Pl Nom stur(er) stur(er) sturu˛ Acc stur(a) stur(er) sturu˛ Dat sturum sturum sturum Though I have not boxed them, some of the systematic syncretisms seen in the noun paradigms are present here: the neuter nominative is always identical to the neuter accusative, in the singular and the plural; the feminine always collapses nominative and accusative in the plural; and the dative plurals are identical in all genders. The long forms (with the material in parentheses) are generally only used in predicative contexts and where the noun is elliptical, for example as follows. (100) a. Gok-etter ien sturo. go-after a big.n.sg.dat ‘Go get a big one’ b. Ig ar tjyöpt diem garu˛. I have bought them large.n.pl.acc ‘I bought them [clothes] large’ c. Bar ig edd dug˚aD fo˛ an frekan. only I had managed get him nice.m.sg.acc ‘If only I could make him nice’ When adjectives are used attributively with an overt noun, they are normally compounded (Delsing 2003), in which case endings consisting of nonnasal vowels with or without /r/ are absent, i.e. the shorter versions of the forms in (99) are used. (101) a. Amm sakt mikkel finkuller jär i bymm. have.1pl truly many fine.girls here in town.def.dat ‘We truly have many fine girls here in town’ b. Ev juo˛t ien kroppknaiv aD mig. take here a curved.knife to me ‘Bring me a curved knife’

39 c. Du ar so stuttan rukk. you have so short.m.acc.sg jacket ‘You have such a short jacket’ Attributive adjectives can be seen in Traditional Ovdalian¨ examples already provided, for example ien sturan krippuop ‘a big.acc child-bunch’ in (14a), or styerr anliening ‘greater reason’ in (62), og iet wiselt fjo˛s ‘a shabby.n barn’ in (75). These show a neuter singular ending -t and occasionally other gender-number endings; they may also show a dative plural -um but other case endings are rare. The form sturan in (14a), from KK, is at least historically distinctively masculine singular accusative, but Levander (1909:107) notes for Classical Ovdalian¨ that this form (in -an) was already then turning up in contexts where other forms could be expected. In general, I have sifted through too few forms to be confident about the pattern in Traditional Ovdalian.¨ This is partly because it is so typical for attributive adjectives to be compounded with the noun they modify, without inflection, especially definite nouns. This can be seen in is-jär sturrottur ‘these big rats’ in (71), or in nog liuotkräld ‘some nasty vessel’ in (89). An example with two adjectives can be seen in (102), from Nyström (1964). (102) Og an-dar sturgr˚arattjin Oskar biet ijel oller trjär! and that-there big.grey.dog.def Oscar bit to.death all three ‘And that big grey dog of Oscar’s bit all three [rats] to death!’ In general, it seems that dative plural outlasts the other case distinctions in adnominal modifiers, unsurprisingly given the state of the nominative-accusative distinction generally in full noun phrases. Dative case is still frequently (and regularly in KK ) distinguished on demonstratives, as illustrated below. The example in (103) shows two examples in direct case, a freestanding demonstrative (eD-dar in standard orthography, neuter) and a prenominal demonstrative (an-dar, masculine). The first of these would historically have been in the dative (dyö-dar), due to the sense of ywyr ‘over’ used here; but recall from section §2.2 that case distinctions in the pronominal system are somewhat out of step with developments in the system of full nouns. (104) gives an example of the dative case (also masculine). (103) Men prester wäsnes uwljuot ywyr e’ dar og stuod i upo˛ ollu but priests wailed terribly over that there and stood in upon all wis fe’ ta fesyets fo˛ brott an dar uosidn. ways for to try get away that there bad.habit.def ‘But the priests wailed terribly about that and made all kinds of efforts to eliminate that immoral behavior’ (104) E’ wa’ so, att kunundsin add ie klucka (an kold o˛n fe’ it was so that the.king had a clock (he called her.acc for byxsäcksur, og wa ljuotouwanliger) i om da’ trouser.pocket.clock and was extremely.unusual) in that.dat there kuppam. the.sack.dat ‘The fact was that the king had a clock (he called it his trouser-pocket clock, and it was extremely unusual) in that sack’

40 Historically, the demonstratives are transparently based on the third person pronominal system, and show the same pattern of syncretism: accusative syn- cretizes with nominative in the singular, but with dative in the plural. The table in (105) shows the paradigm for Classical Ovdalian¨ distal demonstratives. (105) Classical Ovdalian¨ M F N Pl Nom an-dar o-dar eD-dar dier-dar Acc an-dar o-dar eD-dar diem-dar Dat om-dar en-dar dyö-dar diem-dar In the singular, there is no tension between the third person pronominal system and the nominal system, since both collapse the nominative and accusative into direct case (even for the feminine singular, in the demonstratives). But in the plural, there is a tension. As we have seen, plural pronouns in the variety described here make a nominative-objective distinction, while the system of case on full noun phrases makes a direct-dative distinction. The question is then, in the change to a two-case system, what happens to the plural demonstratives? Do they preserve the pronominal differentiation between nominative and objective, or do they realign to conform to the system of the lexical nouns? The answer is that demonstratives (and as far as I can tell, other nominal modifiers) adapt to the new direct-dative system for lexical nouns, with old nominative forms supplanting old objective forms in accusative contexts. Thus, for example, in KK we find examples like the following. The example in (106) is unambiguously nominative, the example in (107) is unambiguously dative on D (though not on N), and the accusative form in (108) (min in this context is expected to take accusative) shows the nominative-like form, not the dative-like form. (106) Dier dar snapperer, wa mjäst werr eld dansker, og ed those there Danish.partisans were mostly worse than Danes and it j˚apt it ur kunundsin fesyekt skrem diem. helped not how the.king tried frighten them ‘Those Danish partisans were mostly worse than Danes, and it didn’t matter how the king tried to frighten them’ (107) E’ wa’ bo’ kripper auti diem dar snapperer it was both children out.in those.dat there Danish.partisans milmad og. sometimes also ‘There were sometimes also children among those Danish partisans’ (108) Nu war ed it gral so ienkelt min dier dar ritualer, sos now was it not completely so simple with those there rituals as ed l˚at. it sounds ‘Now, those rituals weren’t quite as simple as they sound’ The general changeover from a pronominal Nominative-Objective system to a Direct-Dative system in the demonstratives is evidence for the distinction between the two systems, the one operative for pronouns and the one operative for the system of DPs.

41 4 Conclusion

In this paper I have described a case system used by some speakers of Ovdalian,¨ which to my knowledge has not been documented before. The system distinguishes direct from dative case in full noun phrases, and additionally distinguishes nominative from objective cases in the pronominal system, so as a whole it is a three-case system. There are two significant differences between this system and the one described in Akerberg˚ (2000) and Dahl and Koptjevskaja-Tamm (2006). One is the fusion of the nominative and accusative cases, and the other is the fusion of the plural definite and indefinite forms. The latter suggests an analysis in which definite and plural affixes compete for a position. In addition, the near absence of dative case endings when the noun is neither definite nor plural suggests that the contribution of suffixes in the singular has changed compared to earlier stages of the language. There has been a recent resurgence of interest in Ovdalian,¨ and many members of the community have expressed an interest in learning the language, in some cases those whose grandparents or other relatives speak it but who for one reason or another have grown up with Swedish in the home. Fortunately, thanks to the efforts of Ulum Dalska and various enthusiasts, there are now readers, dictionaries, and grammars available for the language. It should be noted, however, that many of the materials available are based on Classical Ovdalian,¨ and do not very accurately reflect the modern speech community, not even the speech of its eldest members. Thus, there might be a place for Traditional Ovdalian¨ in the revitalization efforts that are underway, but before that place can be found, the language itself will have to be better understood. I hope that this paper can make some modest contribution to that end.

References

Afarli,˚ Tor A. and Eldfrid Haaker Fjøsne. 2012. Weak dative case in Norwegian dialect syntax. Studia Linguistica 66 2: 75–93. Akerberg,˚ Bengt. 2000. Ulum Dalska Grammatik 2000 . Ulum Dalska, www.ehf.w.se/Ulum Dalska/. Akerberg,˚ Bengt. 2012. Alvdalsk¨ grammatik. Centrumtryck, Mora. With Gun- nar Nystr¨om. Baerman, Matthew. 2009. Case syncretism. In The Oxford Handbook of Case, edited by Andrej Malchukov and Andrew Spencer. Oxford University Press, Oxford. Bandle, Oskar, Kurt Braunm¨uller, Ernst H˚akon Jahr, Allan Karker, Hans-Peter Naumann, and Ulf Teleman (eds.) . 2002. The Nordic Languages: An Inter- national Handbook of the History of the North Germanic Languages, vol. 1. Walter De Gruyter, Berlin. Blake, Barry J. 2001. Case. Cambridge University Press, Cambridge. Bobaljik, Jonathan David and Dianne Jonas. 1996. Subject positions and the roles of TP. Linguistic Inquiry 27 2: 195–236. Bonet, Eulalia. 1991. Morphology after Syntax: Pronominal Clitics in Romance. Ph.D. thesis, MIT, Cambridge, Ma.

42 Borer, Hagit. 2005. In Name Only (Structuring Sense, vol. I). Oxford University Press, Oxford. Bossong, Georg. 1991. Differential object marking in Romance and beyond. In New Analyses in Romance Linguistics: Selected Papers from the XVIII Linguistic Symposium on Romance Languages, edited by Dieter Wanner and Douglas A. Kibbee, pp. 143–170. John Benjamins, Amsterdam. Bye, Patrik and Peter Svenonius. 2012. Non-concatenative morphology as epiphenomenon. In The Morphology and Phonology of Exponence, edited by Jochen Trommer, pp. 427–495. Oxford University Press, Oxford. Caha, Pavel. 2009. The Nanosyntax of Case. Ph.D. thesis, University of Tromsø. Carstairs, Andrew. 1987. Allomorphy in Inflexion. Croom Helm, London. Dahl, Osten¨ and Maria Koptjevskaja-Tamm. 2006. The resilient dative and other remarkable cases in Scandinavian vernaculars. Sprachtypologie und Uni- versalienforschung (STUF) 59 1: 56–75. Dalrymple, Mary and Irina Nikolaeva. 2011. Objects and Information Structure. Cambridge University Press, Cambridge. DeLancey, Scott. 1981. An interpretation of split ergativity and related patterns. Language 57 3: 626–657. Delsing, Lars-Olof. 2002. The morphology of Old Nordic II: Old Swedish and Old Danish. In The Nordic Languages: An International Handbook of the History of the North Germanic Languages, edited by Oskar Bandle, Kurt Braunmüller, Ernst H˚akon Jahr, Allan Karker, Hans-Peter Naumann, and Ulf Teleman, vol. 1, pp. 925–939. Walter de Gruyter, Berlin. Delsing, Lars-Olof. 2003. Syntaktisk variation i nordiska nominalfraser. In Dialektsyntaktiska studier av den nordiska nominalfrasen, edited by Øys- tein Alexander Vangsnes, Anders Holmberg, and Lars-Olof Delsing, pp. 11–64. Novus, Oslo. Dixon, R. M. W. 1972. The Dyirbal Language of North Queensland. Cambridge University Press, Cambridge. Dixon, R. M. W. 1994. Ergativity. Cambridge University Press, Cambridge. Eythórsson, Thórhallur, Janne Bondi Johannessen, Signe Laake, and Tor A. Afarli.˚ 2012. Dative case in Norwegian, Icelandic and Faroese: Preservation and non-preservation. Journal of Nordic Linguistics 35 3: 219–249. Faarlund, Jan Terje. 2004. The Syntax of Old Norse. Oxford University Press, Oxford. Faarlund, Jan Terje. 2009. On the history of definiteness marking in Scandina- vian. Journal of Linguistics 45 3: 617–639. Garbacz, Piotr. 2010. Word Order in Ovdalian:¨ A Study in Variation and Change. Ph.D. thesis, Lund University. Garbacz, Piotr and Janne Bondi Johannessen. this volume. Ovdalian¨ from 1909 to 2009. In Studies in Ovdalian¨ Syntax, edited by Kristine Bentzen, Henrik Rosenkvist, and Janne Bondi Johannessen. John Benjamins, Amsterdam. Halle, Morris. 1997. Distributed morphology: Impoverishment and fission. MIT Working Papers in Linguistics 30: 425–449. Haspelmath, Martin. 2009. Terminology of case. In The Oxford Handbook of Case, edited by Andrej Malchukov and Andrew Spencer, pp. 505–517. Oxford University Press, Oxford.

43 Helgander, John. 1996. Mobilitet och spr˚akförändring. Exemplet Ovre¨ Dalarna och det vidare perspektivet. Falun, Högskolan Dalarna. Rapport 1996:3. Helgander, John. 2005. Alvdalsm˚al¨ i förändring – n˚agra reflektioner kring en fallstudie. In Rapport fr˚an Fuost konferensn um övdalsku˛ /Första kon- ferensen om älvdalska, edited by Gunnar Nyström. Uppsala Universitet, http://www.nordiska.uu.se/arkiv/konferenser/alvdalska/konferensrapport.htm. Hjelmslev, Louis. 1935. La catégorie des cas: étude de grammaire générale, vol. 1. Universitetsforlaget, Aarhus. Iversen, Ragnvald. 1922. Norrøn grammatikk. Aschehoug, Kristiania. 7th edition 1973. Jakobson, Roman. 1936. Beitrag zur allgemeinen Kasuslehre. Travaux du Cercle Linguistique de Prague 6: 240–288. Lahiri, Aditi, Allison Wetterlin, and Elisabet Jönsson-Steiner. 2005. Sounds definite-ly clitic: Evidence from Scandinavian tone. Lingue e linguaggio 2: 243–262. Larsson, Hjalmar. 1985. Kunundsin kumb: lesubuok o˛ dalska. Published by the author, Alvdalen.¨ Levander, Lars. 1909. Alvdalsm˚alet¨ i Dalarna: Ordböjning ock Syntax. P.A. Norstedt & Söner, Stockholm. Levander, Lars. 1925. Dalm˚alet: Beskrivning och historia, vol. I. Appelberg, Uppsala. Levander, Lars. 1928. Dalm˚alet: Beskrivning och historia. Appelberg, Uppsala. Morén-Duolljá, Bruce. 2013. The prosody of Swedish underived nouns: No lexical tones required. Nordlyd: Tromsø Working Papers in Language and Linguistics 40 1: 196–248. Müller, Gereon. 2005. Syncretism and iconicity in Icelandic noun declensions: A Distributed Morphology approach. Yearbook of Morphology 2004 2004: 229–271. Noreen, Adolf. 1884. Altnordische Grammatik I . Max Niemeyer, Halle. Noreen, Adolf. 1904. Altschwedische Grammatik: mit Einschluss des altgutnis- chen. Max Niemeyer, Halle. Noyer, Ralf. 1992. Features, Positions and Affixes in Autonomous Morphological Structure. Ph.D. thesis, MIT, Cambridge, Ma. Nygaard, Marius. 1966. Norrøn syntax. Emil Moestue Aschehoug, Oslo. Nyström, Gunnar. 1964. Ett prov p˚aKarlarvsm˚alet i älvdalen. Utskrift och översättning av bandinspelning jämte kommentarer. Manuscript, SOFI Upp- sala ref. nr 25661. Pylkkänen, Liina. 2008. Introducing Arguments. MIT Press, Cambridge, Ma. Ringmar, Martin. 2005. Alvdalska¨ – en önordisk spr˚akö p˚a fast- landet? In Rapport fr˚an Fuost konferensn um övdalsku˛ /Första kon- ferensen om älvdalska, edited by Gunnar Nyström. Uppsala Universitet, http://www.nordiska.uu.se/arkiv/konferenser/alvdalska/konferensrapport.htm. Sandøy, Helge. 2000. Vitnem˚al fr˚aein dativbrukar. In Dialekter og folkminnen. Hyllningsskrift till Maj Reinhammar den 17 maj 2000 , edited by Lennart Elmevik, pp. 231–239. Sällskapet för svensk dialektologi, Uppsala. Silverstein, Michael. 1976. Hierarchy of features and ergativity. In Grammatical

44 Categories in Australian Languages, edited by R. M. W. Dixon, pp. 112–171. Australian Institute of Aboriginal Studies, Canberra. Steensland, Lars. 2006. Liten älvdalsk–svensk och svensk–älvdalsk ordbok. Ulum Dalska, Alvdalen.¨ Stump, Gregory T. 2001. Inflectional Morphology: A Theory of Paradigm Struc- ture. Cambridge University Press, Cambridge. Svenonius, Peter. 2008. The position of adjectives and other phrasal modifiers in the decomposition of DP. In Adjectives and Adverbs: Syntax, Semantics, and Discourse, edited by Louise McNally and Chris Kennedy, pp. 16–42. Oxford University Press, Oxford. Svenonius, Peter. to appear. Declension class and the Norwegian definite suffix. In The Morpho-Syntax Phonology Connection, edited by Vera Gribanova and Stephanie Shih, pp. 000–000. Oxford University Press, Oxford. Thráinsson, Höskuldur, Hjalmar P. Petersen, Jógvan ´ıLon Jacobsen, and Za- karis Svabo Hansen. 2004. Faroese: An Overview and Reference Grammar. Føroya FróDskaparfelag, Tórshavn.